Compare commits

..

11 Commits

Author SHA1 Message Date
liangshuo-1
f0b6f35fee fix: resolve schema against runtime metadata in plugin builds; gate cache overlay by version (#1764)
* fix(schema): fall back to runtime catalog when no embedded metadata

Binaries built from the bare Go module (plugin builds) embed only the
empty meta_data_default.json stub because meta_data.json is gitignored
and fetched at build time. The schema command, its completion, and the
affordance command-form resolver read the embedded-only catalog, so
every schema lookup failed with "Unknown service" even though the
runtime registry had already sync-fetched full metadata.

Add registry.SchemaCatalog(): embedded when compiled in (official
builds unchanged, still deterministic), otherwise the merged runtime
catalog seeded from cache or remote fetch. When neither source has
data (offline plugin build with a cold cache), schema now returns a
failed_precondition error with an actionable hint instead of
"Unknown service" with an empty candidate list.

* fix(registry): gate cached meta overlay on version newer than embedded

The cached remote meta was overlaid onto the embedded meta_data.json
unconditionally, so after a CLI upgrade an equal- or older-version
cache kept shadowing the freshly shipped embedded definitions until a
later refresh happened to rewrite it.

Only overlay when the cache version is strictly newer than the
embedded baseline. The bare-module stub baseline is "0.0.0", so plugin
builds without compiled metadata still take any real cached version
(TestOverlayGate_StubEmbedded_OverlaysRealCache) and the schema
runtime fallback keeps working offline from a warm cache.

Ports #1376 onto the typed meta model.

---------

Co-authored-by: liangshuo-1 <266696938+liangshuo-1@users.noreply.github.com>
2026-07-06 21:24:24 +08:00
liangshuo-1
91d785f92f refactor(envvars): consolidate agent env value access (#1757)
* feat(envvars): add LARKSUITE_CLI_AGENT_NAME accessor

envvars.AgentName() reads LARKSUITE_CLI_AGENT_NAME so business code can identify the agent driving the CLI. Value is sanitised (trim, reject control chars, 128-byte cap) and returns "" when unset or invalid. Local-only; not sent to the server.

* refactor(envvars): consolidate agent env accessors, dedupe sanitize

Move agent-trace reading from cmdutil.AgentTraceValue into envvars.AgentTrace(), symmetric with AgentName and sharing one sanitizeSingleLine. secheader.go now only assembles headers (BaseSecurityHeaders calls envvars.AgentTrace()); the duplicated sanitize and its os/unicode imports are gone. The X-Agent-Trace header and its behaviour are unchanged.

---------

Co-authored-by: liangshuo-1 <266696938+liangshuo-1@users.noreply.github.com>
2026-07-06 21:17:51 +08:00
leave330
e621c6e50f fix: support pnpm global installs in self-update (#1705) 2026-07-06 21:11:01 +08:00
Zee Zheng
869a259d4e docs: clarify success envelope contract — judge success by ok, not code (#1730)
The stdout success envelope is {ok, identity, data, meta?} and carries no
top-level code/msg field, but wrappers following the raw OpenAPI convention
test code == 0 and misclassify every successful call as a failure. Around
write commands this defeats caller-side idempotency and causes duplicate
creates (see #1689).

Document the contract in ERROR_CONTRACT.md (success-envelope counterpart to
the error wire format), both READMEs (JSON Output Contract section), the
lark-shared skill (shared prerequisite for all skills), and add a success
response example to lark-task-create.md.

Closes #1689
2026-07-06 20:03:05 +08:00
fangshuyu-768
ee46e22abd docs: tighten doc creation validation workflow (#1759) 2026-07-06 17:07:10 +08:00
hhang
b76dc18c2f fix(calendar): guide approval room booking fallback (#1637) 2026-07-06 15:04:50 +08:00
calendar-assistant
85679d4258 feat: support semantic recurring calendar operations (#1723) 2026-07-06 14:10:57 +08:00
wangweiming-01
1ba4f3973c fix: guide drive import concurrency conflicts (#1751) 2026-07-06 10:46:46 +08:00
liangshuo-1
c45ff569c4 chore: release v1.0.65 (#1742) 2026-07-03 20:25:22 +08:00
caojie0621
a1506cdffb feat: add docs history shortcuts (#1612)
Add docs +history-list, +history-revert, and +history-revert-status backed by docs_ai history OpenAPI endpoints.

Document the safe history workflow and extend dry-run/live E2E coverage for the new shortcuts.
2026-07-03 16:21:18 +08:00
liuxin-0319
3595356ea1 chore: sync lark-doc skill from online-doc (#1701) 2026-07-03 15:55:46 +08:00
60 changed files with 2555 additions and 648 deletions

3
.gitignore vendored
View File

@@ -27,6 +27,9 @@ Thumbs.db
# Go
docs/ref
docs/
!tests/cli_e2e/docs/
!tests/cli_e2e/docs/*.go
!tests/cli_e2e/docs/*.md
vendor/

View File

@@ -2,6 +2,22 @@
All notable changes to this project will be documented in this file.
## [v1.0.65] - 2026-07-03
### Features
- **doc**: Add `+history-list`, `+history-revert`, and `+history-revert-status` shortcuts for document version history (#1612)
### Bug Fixes
- **minutes**: `+speaker-replace` no longer refetches the speaker list — `--from-speaker-id` is passed through as-is (#1731)
### Documentation
- **drive**: Document 30-char query limit for `+search` (#1560)
- **doc**: Add mindnote guidance to lark-doc skill (#1581)
- **doc**: Sync lark-doc skill content from online-doc (#1701)
## [v1.0.64] - 2026-07-02
### Features
@@ -1355,6 +1371,7 @@ Bundled AI agent skills for intelligent assistance:
- Bilingual documentation (English & Chinese).
- CI/CD pipelines: linting, testing, coverage reporting, and automated releases.
[v1.0.65]: https://github.com/larksuite/cli/releases/tag/v1.0.65
[v1.0.64]: https://github.com/larksuite/cli/releases/tag/v1.0.64
[v1.0.62]: https://github.com/larksuite/cli/releases/tag/v1.0.62
[v1.0.61]: https://github.com/larksuite/cli/releases/tag/v1.0.61

View File

@@ -233,6 +233,24 @@ lark-cli api POST /open-apis/im/v1/messages --params '{"receive_id_type":"chat_i
--format csv # Comma-separated values
```
### JSON Output Contract
With `--format json` (the default), success and error envelopes are distinct.
Success goes to **stdout**, exit code `0`:
```json
{ "ok": true, "identity": "user", "data": { "guid": "..." }, "meta": { "count": 1 } }
```
Errors go to **stderr**, non-zero exit code:
```json
{ "ok": false, "identity": "user", "error": { "type": "api", "subtype": "...", "code": 99991679, "message": "...", "hint": "..." } }
```
To check whether a command succeeded, test `ok == true` (or the exit code) — **not** `code == 0`. Unlike raw OpenAPI responses (`{"code": 0, "msg": "ok", ...}`), the success envelope carries no `code` or `msg` field; `code` appears only inside `error` as the upstream OpenAPI code. See [errs/ERROR_CONTRACT.md](errs/ERROR_CONTRACT.md) for the full error taxonomy.
### Pagination
```bash

View File

@@ -234,6 +234,24 @@ lark-cli api POST /open-apis/im/v1/messages --params '{"receive_id_type":"chat_i
--format csv # 逗号分隔值
```
### JSON 输出契约
`--format json`(默认)下,成功与错误的信封结构不同。
成功信封写入 **stdout**,退出码 0
```json
{ "ok": true, "identity": "user", "data": { "guid": "..." }, "meta": { "count": 1 } }
```
错误信封写入 **stderr**,退出码非 0
```json
{ "ok": false, "identity": "user", "error": { "type": "api", "subtype": "...", "code": 99991679, "message": "...", "hint": "..." } }
```
判断命令是否成功,请检查 `ok == true`(或进程退出码),**不要用 `code == 0`**。与原始 OpenAPI 响应(`{"code": 0, "msg": "ok", ...}`)不同,成功信封没有 `code``msg` 字段;`code` 只出现在错误信封的 `error` 内,含义是上游 OpenAPI 的 numeric code。完整错误分类见 [errs/ERROR_CONTRACT.md](errs/ERROR_CONTRACT.md)。
### 分页
```bash

View File

@@ -20,13 +20,28 @@ import (
"github.com/spf13/cobra"
)
func newTestApiCmd(f *cmdutil.Factory, runF func(*APIOptions) error) *cobra.Command {
cmd := NewCmdApi(f, runF)
cmd.SilenceErrors = true
cmd.SilenceUsage = true
return cmd
}
func newTestRootCmd() *cobra.Command {
return &cobra.Command{
Use: "lark-cli",
SilenceErrors: true,
SilenceUsage: true,
}
}
func TestApiCmd_FlagParsing(t *testing.T) {
f, _, _, _ := cmdutil.TestFactory(t, &core.CliConfig{
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
var gotOpts *APIOptions
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
gotOpts = opts
return nil
})
@@ -54,7 +69,7 @@ func TestApiCmd_DryRun(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/test", "--as", "bot", "--dry-run"})
err := cmd.Execute()
if err != nil {
@@ -77,7 +92,7 @@ func TestApiCmd_NullParamsWithPageSize(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/test", "--params", "null", "--page-size", "50", "--as", "bot", "--dry-run"})
if err := cmd.Execute(); err != nil {
t.Fatalf("--params null with --page-size should not error, got: %v", err)
@@ -98,7 +113,7 @@ func TestApiCmd_BotMode(t *testing.T) {
Body: map[string]interface{}{"code": 0, "msg": "ok", "data": map[string]interface{}{"result": "success"}},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/test", "--as", "bot"})
err := cmd.Execute()
if err != nil {
@@ -125,7 +140,7 @@ func TestApiCmd_MissingArgs(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET"}) // missing path
err := cmd.Execute()
if err == nil {
@@ -138,7 +153,7 @@ func TestApiCmd_InvalidParamsJSON(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/test", "--as", "bot", "--params", "{bad"})
err := cmd.Execute()
if err == nil {
@@ -151,7 +166,7 @@ func TestApiValidArgsFunction(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
fn := cmd.ValidArgsFunction
tests := []struct {
@@ -217,7 +232,7 @@ func TestNewCmdApi_StrictModeHidesAsFlag(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu, SupportedIdentities: 2,
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
flag := cmd.Flags().Lookup("as")
if flag == nil {
t.Fatal("expected --as flag to be registered")
@@ -236,7 +251,7 @@ func TestApiCmd_PageLimitDefault(t *testing.T) {
})
var gotOpts *APIOptions
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
gotOpts = opts
return nil
})
@@ -255,7 +270,7 @@ func TestApiCmd_ParamsAndDataBothStdinConflict(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"POST", "/open-apis/test", "--as", "bot", "--params", "-", "--data", "-"})
err := cmd.Execute()
if err == nil {
@@ -272,7 +287,7 @@ func TestApiCmd_OutputAndPageAllConflict(t *testing.T) {
})
var gotOpts *APIOptions
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
gotOpts = opts
return apiRun(opts)
})
@@ -297,7 +312,7 @@ func TestApiCmd_BinaryResponse_AutoSave(t *testing.T) {
ContentType: "application/octet-stream",
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/drive/v1/files/xxx/download", "--as", "bot"})
err := cmd.Execute()
if err != nil {
@@ -328,7 +343,7 @@ func TestApiCmd_PageAll_NonBatchAPI_FallbackToJSON(t *testing.T) {
},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/contact/v3/users/u123", "--as", "bot", "--page-all", "--format", "ndjson"})
err := cmd.Execute()
if err != nil {
@@ -368,7 +383,7 @@ func TestApiCmd_PageAll_NonBatchAPI_ErrorStillOutputsJSON(t *testing.T) {
},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/im/v1/chats/oc_xxx/announcement", "--as", "bot", "--page-all"})
err := cmd.Execute()
// Should return an error
@@ -409,7 +424,7 @@ func TestApiCmd_PageAll_BatchAPI_StreamsItems(t *testing.T) {
},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/contact/v3/users", "--as", "bot", "--page-all", "--format", "ndjson"})
err := cmd.Execute()
if err != nil {
@@ -448,7 +463,7 @@ func TestApiCmd_PageAll_StreamBusinessErrorDoesNotDumpJSON(t *testing.T) {
},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/contact/v3/users", "--as", "bot", "--page-all", "--format", "ndjson"})
err := cmd.Execute()
if err == nil {
@@ -483,7 +498,7 @@ func TestApiCmd_PageAll_BatchAPI_DefaultJSONEnvelope(t *testing.T) {
},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/contact/v3/users", "--as", "bot", "--page-all"})
if err := cmd.Execute(); err != nil {
t.Fatalf("unexpected error: %v", err)
@@ -549,8 +564,8 @@ func TestApiCmd_PageAll_DefaultJSONRunsContentSafety(t *testing.T) {
},
})
root := &cobra.Command{Use: "lark-cli"}
root.AddCommand(NewCmdApi(f, nil))
root := newTestRootCmd()
root.AddCommand(newTestApiCmd(f, nil))
root.SetArgs([]string{"api", "GET", "/open-apis/contact/v3/users", "--as", "bot", "--page-all"})
if err := root.Execute(); err != nil {
t.Fatalf("unexpected error: %v", err)
@@ -600,8 +615,8 @@ func TestApiCmd_PageAll_StreamFormatRunsContentSafety(t *testing.T) {
},
})
root := &cobra.Command{Use: "lark-cli"}
root.AddCommand(NewCmdApi(f, nil))
root := newTestRootCmd()
root.AddCommand(newTestApiCmd(f, nil))
root.SetArgs([]string{"api", "GET", "/open-apis/contact/v3/users", "--as", "bot", "--page-all", "--format", "ndjson"})
if err := root.Execute(); err != nil {
t.Fatalf("unexpected error: %v", err)
@@ -656,8 +671,8 @@ func TestApiCmd_PageAll_StreamFormatBlockSkipsBlockedPage(t *testing.T) {
},
})
root := &cobra.Command{Use: "lark-cli"}
root.AddCommand(NewCmdApi(f, nil))
root := newTestRootCmd()
root.AddCommand(newTestApiCmd(f, nil))
root.SetArgs([]string{"api", "GET", "/open-apis/contact/v3/users", "--as", "bot", "--page-all", "--format", "ndjson"})
err := root.Execute()
if err == nil {
@@ -721,7 +736,7 @@ func TestApiCmd_JqFlag_Parsing(t *testing.T) {
})
var gotOpts *APIOptions
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
gotOpts = opts
return nil
})
@@ -741,7 +756,7 @@ func TestApiCmd_JqFlag_ShortForm(t *testing.T) {
})
var gotOpts *APIOptions
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
gotOpts = opts
return nil
})
@@ -760,7 +775,7 @@ func TestApiCmd_JqAndOutputConflict(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
return apiRun(opts)
})
cmd.SetArgs([]string{"GET", "/open-apis/test", "--as", "bot", "--jq", ".data", "--output", "file.bin"})
@@ -791,7 +806,7 @@ func TestApiCmd_JqFilter_AppliesExpression(t *testing.T) {
},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/test/jq", "--as", "bot", "--jq", ".data.items[].name"})
err := cmd.Execute()
if err != nil {
@@ -812,7 +827,7 @@ func TestApiCmd_JqAndFormatConflict(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
return apiRun(opts)
})
cmd.SetArgs([]string{"GET", "/open-apis/test", "--as", "bot", "--jq", ".data", "--format", "ndjson"})
@@ -830,7 +845,7 @@ func TestApiCmd_JqInvalidExpression(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
return apiRun(opts)
})
cmd.SetArgs([]string{"GET", "/open-apis/test", "--as", "bot", "--jq", "invalid["})
@@ -859,7 +874,7 @@ func TestApiCmd_PageAll_WithJq(t *testing.T) {
},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/contact/v3/users", "--as", "bot", "--page-all", "--jq", ".data.items[].id"})
err := cmd.Execute()
if err != nil {
@@ -880,7 +895,7 @@ func TestApiCmd_MethodUppercase(t *testing.T) {
})
var gotOpts *APIOptions
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
gotOpts = opts
return nil
})
@@ -899,7 +914,7 @@ func TestApiCmd_FileFlagParsing(t *testing.T) {
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
var gotOpts *APIOptions
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
gotOpts = opts
return nil
})
@@ -917,7 +932,7 @@ func TestApiCmd_FileAndOutputConflict(t *testing.T) {
f, _, _, _ := cmdutil.TestFactory(t, &core.CliConfig{
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
return apiRun(opts)
})
cmd.SetArgs([]string{"POST", "/open-apis/test", "--as", "bot", "--file", "photo.jpg", "--output", "out.json"})
@@ -934,7 +949,7 @@ func TestApiCmd_FileWithGET(t *testing.T) {
f, _, _, _ := cmdutil.TestFactory(t, &core.CliConfig{
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
return apiRun(opts)
})
cmd.SetArgs([]string{"GET", "/open-apis/test", "--as", "bot", "--file", "photo.jpg"})
@@ -951,7 +966,7 @@ func TestApiCmd_FileStdinConflictWithData(t *testing.T) {
f, _, _, _ := cmdutil.TestFactory(t, &core.CliConfig{
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
return apiRun(opts)
})
cmd.SetArgs([]string{"POST", "/open-apis/test", "--as", "bot", "--file", "-", "--data", "-"})
@@ -974,7 +989,7 @@ func TestApiCmd_DryRunWithFile(t *testing.T) {
f, stdout, _, _ := cmdutil.TestFactory(t, &core.CliConfig{
AppID: "test-app", AppSecret: "test-secret", Brand: core.BrandFeishu,
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"POST", "/open-apis/im/v1/images", "--file", "image=" + tmpFile, "--data", `{"image_type":"message"}`, "--dry-run", "--as", "bot"})
err := cmd.Execute()
if err != nil {
@@ -1015,7 +1030,7 @@ func TestApiCmd_PermissionError_DerivesFirstClassFields(t *testing.T) {
},
})
cmd := NewCmdApi(f, nil)
cmd := newTestApiCmd(f, nil)
cmd.SetArgs([]string{"GET", "/open-apis/docx/v1/documents/test", "--as", "bot"})
err := cmd.Execute()
if err == nil {
@@ -1041,7 +1056,7 @@ func TestApiCmd_JsonFlag_Accepted(t *testing.T) {
})
var gotOpts *APIOptions
cmd := NewCmdApi(f, func(opts *APIOptions) error {
cmd := newTestApiCmd(f, func(opts *APIOptions) error {
gotOpts = opts
return nil
})

View File

@@ -65,13 +65,13 @@ func NewCmdSchema(f *cmdutil.Factory, runF func(*SchemaOptions) error) *cobra.Co
return cmd
}
// completeSchemaPath is a thin adapter over the embedded catalog's Complete.
// It uses the embedded source so completion candidates match what `schema`
// execution can resolve (both overlay-free).
// completeSchemaPath is a thin adapter over the schema catalog's Complete.
// It uses the same source as schema execution so completion candidates match
// what `schema` can resolve.
func completeSchemaPath(f *cmdutil.Factory) func(*cobra.Command, []string, string) ([]string, cobra.ShellCompDirective) {
return func(cmd *cobra.Command, args []string, toComplete string) ([]string, cobra.ShellCompDirective) {
mode := f.ResolveStrictMode(cmd.Context())
completions, noSpace := registry.EmbeddedCatalog().Complete(args, toComplete, registry.FilterForStrictMode(mode))
completions, noSpace := registry.SchemaCatalog().Complete(args, toComplete, registry.FilterForStrictMode(mode))
directive := cobra.ShellCompDirectiveNoFileComp
if noSpace {
directive |= cobra.ShellCompDirectiveNoSpace
@@ -86,13 +86,19 @@ func schemaRun(opts *SchemaOptions) error {
return runSchema(out, apicatalog.ParsePath(opts.Args), mode)
}
// runSchema resolves the path through the embedded catalog and renders the
// runSchema resolves the path through the schema catalog and renders the
// matching envelope(s). The catalog owns navigation (Resolve + MethodRefs) and
// schema owns rendering (Envelope/Envelopes); this adapter only chooses the
// output shape — a single resolved method renders as one envelope object,
// anything broader as an array — and maps resolve failures to hints.
func runSchema(out io.Writer, parts []string, mode core.StrictMode) error {
catalog := registry.EmbeddedCatalog()
catalog := registry.SchemaCatalog()
if len(catalog.Services()) == 0 {
// No embedded metadata and the runtime fallback is empty too: offline
// with a cold cache, remote meta off, or an unwritable cache dir.
return errs.NewValidationError(errs.SubtypeFailedPrecondition, "No API metadata available").
WithHint("this binary has no embedded API metadata; run any command with network access to the open platform once so metadata can be fetched and cached")
}
target, err := catalog.Resolve(parts)
if err != nil {
return resolveError(err)

View File

@@ -102,7 +102,8 @@ func NewCmdUpdate(f *cmdutil.Factory) *cobra.Command {
Long: `Update lark-cli to the latest version.
Detects the installation method automatically:
- npm install: runs npm install -g @larksuite/cli@<version>
- npm install: runs npm install -g @larksuite/cli@<version>
- pnpm install: runs pnpm add -g @larksuite/cli@<version>
- manual/other: shows GitHub Releases download URL
Use --json for structured output (for AI agents and scripts).
@@ -164,7 +165,7 @@ func updateRun(opts *UpdateOptions) error {
if !detect.CanAutoUpdate() {
return doManualUpdate(opts, io, cur, latest, detect, updater)
}
return doNpmUpdate(opts, io, cur, latest, updater)
return doAutoUpdate(opts, io, cur, latest, detect, updater)
}
// --- Output helpers ---
@@ -226,12 +227,23 @@ func doManualUpdate(opts *UpdateOptions, io *cmdutil.IOStreams, cur, latest stri
fmt.Fprintf(io.ErrOut, "To update manually, download the latest release:\n")
fmt.Fprintf(io.ErrOut, " Release: %s\n", releaseURL(latest))
fmt.Fprintf(io.ErrOut, " Changelog: %s\n", changelogURL())
fmt.Fprintf(io.ErrOut, "\nOr install via npm (note: skills will not be synced):\n npm install -g %s@%s\n npx skills add larksuite/cli -y -g # sync skills separately\n", selfupdate.NpmPackage, latest)
if detect.Method == selfupdate.InstallPnpm {
fmt.Fprintf(io.ErrOut, "\nOr install via pnpm (note: skills will not be synced):\n pnpm add -g %s@%s\n pnpm dlx skills add larksuite/cli -y -g # sync skills separately\n", selfupdate.NpmPackage, latest)
} else {
fmt.Fprintf(io.ErrOut, "\nOr install via npm (note: skills will not be synced):\n npm install -g %s@%s\n npx skills add larksuite/cli -y -g # sync skills separately\n", selfupdate.NpmPackage, latest)
}
emitSkillsTextHints(io, skillsResult)
return nil
}
func doNpmUpdate(opts *UpdateOptions, io *cmdutil.IOStreams, cur, latest string, updater *selfupdate.Updater) error {
func doAutoUpdate(opts *UpdateOptions, io *cmdutil.IOStreams, cur, latest string, detect selfupdate.DetectResult, updater *selfupdate.Updater) error {
pm := "npm"
install := updater.RunNpmInstall
if detect.Method == selfupdate.InstallPnpm {
pm = "pnpm"
install = updater.RunPnpmInstall
}
restore, err := updater.PrepareSelfReplace()
if err != nil {
return reportError(opts, io, "update_error",
@@ -239,19 +251,19 @@ func doNpmUpdate(opts *UpdateOptions, io *cmdutil.IOStreams, cur, latest string,
}
if !opts.JSON {
fmt.Fprintf(io.ErrOut, "Updating lark-cli %s %s %s via npm ...\n", cur, symArrow(), latest)
fmt.Fprintf(io.ErrOut, "Updating lark-cli %s %s %s via %s ...\n", cur, symArrow(), latest, pm)
}
npmResult := updater.RunNpmInstall(latest)
npmResult := install(latest)
if npmResult.Err != nil {
restore()
combined := npmResult.CombinedOutput()
if opts.JSON {
output.PrintJson(io.Out, map[string]interface{}{
"ok": false, "error": map[string]interface{}{
"type": "update_error", "message": fmt.Sprintf("npm install failed: %s", npmResult.Err),
"type": "update_error", "message": fmt.Sprintf("%s install failed: %s", pm, npmResult.Err),
"detail": selfupdate.Truncate(combined, maxNpmOutput),
"hint": permissionHint(combined),
"hint": permissionHint(combined, pm),
},
})
return output.ErrBare(output.ExitAPI)
@@ -263,7 +275,7 @@ func doNpmUpdate(opts *UpdateOptions, io *cmdutil.IOStreams, cur, latest string,
fmt.Fprint(io.ErrOut, npmResult.Stderr.String())
}
fmt.Fprintf(io.ErrOut, "\n%s Update failed: %s\n", symFail(), npmResult.Err)
if hint := permissionHint(combined); hint != "" {
if hint := permissionHint(combined, pm); hint != "" {
fmt.Fprintf(io.ErrOut, " %s\n", hint)
}
return output.ErrBare(output.ExitAPI)
@@ -274,7 +286,7 @@ func doNpmUpdate(opts *UpdateOptions, io *cmdutil.IOStreams, cur, latest string,
if err := updater.VerifyBinary(latest); err != nil {
restore()
msg := fmt.Sprintf("new binary verification failed: %s", err)
hint := verificationFailureHint(updater, latest)
hint := verificationFailureHint(updater, latest, pm)
if opts.JSON {
output.PrintJson(io.Out, map[string]interface{}{
"ok": false,
@@ -304,23 +316,33 @@ func doNpmUpdate(opts *UpdateOptions, io *cmdutil.IOStreams, cur, latest string,
fmt.Fprintf(io.ErrOut, "\n%s Successfully updated lark-cli from %s to %s\n", symOK(), cur, latest)
fmt.Fprintf(io.ErrOut, " Changelog: %s\n", changelogURL())
if skillsResult != nil {
fmt.Fprintf(io.ErrOut, "\nUpdating skills ...\n")
skillsPM := "npx"
if detect.Method == selfupdate.InstallPnpm && detect.PnpmAvailable {
skillsPM = "pnpm dlx"
}
fmt.Fprintf(io.ErrOut, "\nUpdating skills via %s ...\n", skillsPM)
}
emitSkillsTextHints(io, skillsResult)
return nil
}
func permissionHint(npmOutput string) string {
if strings.Contains(npmOutput, "EACCES") && !isWindows() {
return "Permission denied. Try: sudo lark-cli update, or adjust your npm global prefix: https://docs.npmjs.com/resolving-eacces-permissions-errors"
func permissionHint(pmOutput, pm string) string {
if !strings.Contains(pmOutput, "EACCES") || isWindows() {
return ""
}
return ""
if pm == "pnpm" {
return "Permission denied. Ensure your pnpm global directory is writable — re-run `pnpm setup`, or see https://pnpm.io/pnpm-cli"
}
return "Permission denied. Try: sudo lark-cli update, or adjust your npm global prefix: https://docs.npmjs.com/resolving-eacces-permissions-errors"
}
func verificationFailureHint(updater *selfupdate.Updater, latest string) string {
func verificationFailureHint(updater *selfupdate.Updater, latest, pm string) string {
if updater.CanRestorePreviousVersion() {
return "the previous version has been restored"
}
if pm == "pnpm" {
return fmt.Sprintf("automatic rollback is unavailable on this platform; reinstall manually (skills will not be synced): pnpm add -g %s@%s && pnpm dlx skills add larksuite/cli -y -g, or download %s", selfupdate.NpmPackage, latest, releaseURL(latest))
}
return fmt.Sprintf("automatic rollback is unavailable on this platform; reinstall manually (skills will not be synced): npm install -g %s@%s && npx skills add larksuite/cli -y -g, or download %s", selfupdate.NpmPackage, latest, releaseURL(latest))
}

View File

@@ -57,6 +57,27 @@ func mockDetectAndNpm(t *testing.T, result selfupdate.DetectResult, npmFn func(s
t.Cleanup(func() { newUpdater = origNew })
}
// mockDetectAndPnpm mirrors mockDetectAndNpm but wires the pnpm install path
// and fails the test if the npm install path is invoked.
func mockDetectAndPnpm(t *testing.T, result selfupdate.DetectResult, pnpmFn func(string) *selfupdate.NpmResult) {
t.Helper()
origNew := newUpdater
newUpdater = func() *selfupdate.Updater {
u := selfupdate.New()
u.DetectOverride = func() selfupdate.DetectResult { return result }
u.PnpmInstallOverride = pnpmFn
u.NpmInstallOverride = func(string) *selfupdate.NpmResult {
t.Errorf("npm install must not be called for a pnpm install")
return &selfupdate.NpmResult{}
}
u.VerifyOverride = func(string) error { return nil }
u.SkillsIndexFetchOverride = successfulSkillsIndexFetch()
u.SkillsCommandOverride = successfulSkillsCommand()
return u
}
t.Cleanup(func() { newUpdater = origNew })
}
func successfulSkillsIndexFetch() func() *selfupdate.NpmResult {
return func() *selfupdate.NpmResult {
r := &selfupdate.NpmResult{}
@@ -81,6 +102,110 @@ func successfulSkillsCommand() func(args ...string) *selfupdate.NpmResult {
}
}
func TestUpdatePnpm_JSON(t *testing.T) {
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", t.TempDir())
f, stdout, _ := newTestFactory(t)
cmd := NewCmdUpdate(f)
cmd.SetArgs([]string{"--json"})
origFetch := fetchLatest
fetchLatest = func() (string, error) { return "2.0.0", nil }
defer func() { fetchLatest = origFetch }()
origVersion := currentVersion
currentVersion = func() string { return "1.0.0" }
defer func() { currentVersion = origVersion }()
mockDetectAndPnpm(t,
selfupdate.DetectResult{Method: selfupdate.InstallPnpm, ResolvedPath: "/x/node_modules/.pnpm/@larksuite+cli@1.0.0/node_modules/@larksuite/cli/bin/lark-cli", PnpmAvailable: true},
func(string) *selfupdate.NpmResult { return &selfupdate.NpmResult{} },
)
if err := cmd.Execute(); err != nil {
t.Fatalf("unexpected error: %v", err)
}
if out := stdout.String(); !strings.Contains(out, `"action": "updated"`) {
t.Errorf("expected updated in output, got: %s", out)
}
}
func TestUpdatePnpm_Human(t *testing.T) {
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", t.TempDir())
f, _, stderr := newTestFactory(t)
cmd := NewCmdUpdate(f)
cmd.SetArgs([]string{})
origFetch := fetchLatest
fetchLatest = func() (string, error) { return "2.0.0", nil }
defer func() { fetchLatest = origFetch }()
origVersion := currentVersion
currentVersion = func() string { return "1.0.0" }
defer func() { currentVersion = origVersion }()
mockDetectAndPnpm(t,
selfupdate.DetectResult{Method: selfupdate.InstallPnpm, ResolvedPath: "/x/node_modules/.pnpm/@larksuite+cli@1.0.0/node_modules/@larksuite/cli/bin/lark-cli", PnpmAvailable: true},
func(string) *selfupdate.NpmResult { return &selfupdate.NpmResult{} },
)
if err := cmd.Execute(); err != nil {
t.Fatalf("unexpected error: %v", err)
}
out := stderr.String()
if !strings.Contains(out, "via pnpm") {
t.Errorf("expected 'via pnpm' in stderr, got: %s", out)
}
if !strings.Contains(out, "Updating skills via pnpm dlx ...") {
t.Errorf("expected skills sync to report pnpm dlx launcher, got: %s", out)
}
if !strings.Contains(out, "Successfully updated") {
t.Errorf("expected success message, got: %s", out)
}
}
func TestUpdatePnpm_InstallError_JSON(t *testing.T) {
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", t.TempDir())
f, stdout, _ := newTestFactory(t)
cmd := NewCmdUpdate(f)
cmd.SetArgs([]string{"--json"})
origFetch := fetchLatest
fetchLatest = func() (string, error) { return "2.0.0", nil }
defer func() { fetchLatest = origFetch }()
origVersion := currentVersion
currentVersion = func() string { return "1.0.0" }
defer func() { currentVersion = origVersion }()
mockDetectAndPnpm(t,
selfupdate.DetectResult{Method: selfupdate.InstallPnpm, ResolvedPath: "/x/node_modules/.pnpm/@larksuite+cli@1.0.0/node_modules/@larksuite/cli/bin/lark-cli", PnpmAvailable: true},
func(string) *selfupdate.NpmResult { return &selfupdate.NpmResult{Err: errors.New("pnpm boom")} },
)
err := cmd.Execute()
if err == nil {
t.Fatal("expected error exit")
}
if out := stdout.String(); !strings.Contains(out, `"ok": false`) || !strings.Contains(out, "update_error") {
t.Errorf("expected failure envelope, got: %s", out)
}
if out := stdout.String(); !strings.Contains(out, "pnpm install failed") {
t.Errorf("expected message to report pnpm as the package manager, got: %s", out)
}
}
func TestUpdatePnpm_Unavailable_ManualFallback(t *testing.T) {
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", t.TempDir())
f, _, stderr := newTestFactory(t)
cmd := NewCmdUpdate(f)
cmd.SetArgs([]string{})
origFetch := fetchLatest
fetchLatest = func() (string, error) { return "2.0.0", nil }
defer func() { fetchLatest = origFetch }()
origVersion := currentVersion
currentVersion = func() string { return "1.0.0" }
defer func() { currentVersion = origVersion }()
mockDetect(t, selfupdate.DetectResult{Method: selfupdate.InstallPnpm, ResolvedPath: "/x/node_modules/.pnpm/@larksuite+cli@1.0.0/node_modules/@larksuite/cli/bin/lark-cli", PnpmAvailable: false})
if err := cmd.Execute(); err != nil {
t.Fatalf("unexpected error: %v", err)
}
out := stderr.String()
if !strings.Contains(out, "installed via pnpm, but pnpm is not available in PATH") {
t.Errorf("expected pnpm manual reason, got: %s", out)
}
if !strings.Contains(out, "pnpm add -g") {
t.Errorf("expected pnpm add -g hint, got: %s", out)
}
}
func TestNormalizeVersion(t *testing.T) {
tests := []struct {
input string
@@ -266,6 +391,9 @@ func TestUpdateNpm_Human(t *testing.T) {
if !strings.Contains(out, "Successfully updated") {
t.Errorf("expected success message in stderr, got: %s", out)
}
if !strings.Contains(out, "Updating skills via npx ...") {
t.Errorf("expected skills sync to report npx launcher for npm install, got: %s", out)
}
}
func TestUpdateForce_JSON(t *testing.T) {
@@ -739,9 +867,9 @@ func TestPermissionHint(t *testing.T) {
origOS := currentOS
defer func() { currentOS = origOS }()
// Linux: EACCES should produce a hint with npm prefix guidance.
// Linux + npm: EACCES should produce a hint with npm prefix guidance.
currentOS = "linux"
hint := permissionHint("EACCES: permission denied, access '/usr/local/lib'")
hint := permissionHint("EACCES: permission denied, access '/usr/local/lib'", "npm")
if !strings.Contains(hint, "npm global prefix") {
t.Errorf("expected npm prefix hint on linux, got: %s", hint)
}
@@ -749,16 +877,25 @@ func TestPermissionHint(t *testing.T) {
t.Errorf("should not suggest raw sudo npm install, got: %s", hint)
}
// Linux + pnpm: EACCES should point at pnpm setup, not npm prefix/sudo.
pnpmHint := permissionHint("EACCES: permission denied, access '/Users/x/Library/pnpm'", "pnpm")
if !strings.Contains(pnpmHint, "pnpm setup") {
t.Errorf("expected pnpm setup hint, got: %s", pnpmHint)
}
if strings.Contains(pnpmHint, "npm global prefix") || strings.Contains(pnpmHint, "sudo") {
t.Errorf("pnpm hint must not reference npm prefix or sudo, got: %s", pnpmHint)
}
// Windows: EACCES hint is suppressed (no EACCES on Windows).
currentOS = "windows"
hint = permissionHint("EACCES: permission denied")
hint = permissionHint("EACCES: permission denied", "npm")
if hint != "" {
t.Errorf("expected empty hint on Windows, got: %s", hint)
}
// Non-EACCES error: always empty.
currentOS = "linux"
if got := permissionHint("some other error"); got != "" {
if got := permissionHint("some other error", "npm"); got != "" {
t.Errorf("expected empty hint for non-EACCES, got: %s", got)
}
}

View File

@@ -72,6 +72,28 @@ other category. `error.type` is `"policy"`, `error.subtype` is one of
`challenge_required` / `access_denied`, and process exit is `6` via
`CategoryPolicy`.
### Success envelope (stdout)
For contrast: success responses render to **stdout** as an
`output.Envelope` (`internal/output/envelope.go`), exit code `0`:
```json
{
"ok": true,
"identity": "user",
"data": { "guid": "e297d3d0-..." },
"meta": { "count": 1 }
}
```
Consumers must branch on `ok` (or the process exit code). The success
envelope has **no top-level `code` or `msg` field**`code` exists only
inside `error`, where it is the upstream numeric code (invariant 4).
Wrappers that follow the raw OpenAPI convention and test `code == 0`
will misclassify every successful call as a failure, which is
especially dangerous around write commands (e.g. retrying a create that
already succeeded).
## Categories
| Category | When | Exit | Typed struct |

View File

@@ -77,14 +77,10 @@ func loadService(service string) map[string]json.RawMessage {
// space→dot fallback covers domains where the two already coincide.
func commandFormResolver(service string) func(string) string {
byForm := map[string]string{}
for _, svc := range registry.EmbeddedServicesTyped() {
if svc.Name != service {
continue
}
if svc, ok := registry.SchemaCatalog().Service(service); ok {
for _, ref := range apicatalog.ServiceMethods(svc, nil) {
byForm[strings.Join(ref.CommandPath()[1:], " ")] = ref.Method.ID
}
break
}
return func(h string) string {
h = strings.TrimSpace(h)

View File

@@ -6,12 +6,10 @@ package cmdutil
import (
"context"
"net/http"
"os"
"reflect"
"runtime/debug"
"strings"
"sync"
"unicode"
"github.com/larksuite/cli/extension/credential"
"github.com/larksuite/cli/extension/fileio"
@@ -40,8 +38,6 @@ const (
BuildKindUnknown = "unknown"
officialModulePath = "github.com/larksuite/cli"
agentTraceMaxLen = 1024
)
// UserAgentValue returns the User-Agent value: "lark-cli/{version}".
@@ -49,25 +45,6 @@ func UserAgentValue() string {
return SourceValue + "/" + build.Version
}
// AgentTraceValue returns a header-safe value from the
// LARKSUITE_CLI_AGENT_TRACE environment variable. It trims
// surrounding whitespace, rejects values containing any Unicode
// control character or exceeding agentTraceMaxLen, and returns ""
// for any invalid or empty value. Callers can use the result
// directly in HTTP headers without further sanitisation.
func AgentTraceValue() string {
v := strings.TrimSpace(os.Getenv(envvars.CliAgentTrace))
if v == "" || len(v) > agentTraceMaxLen {
return ""
}
for _, r := range v {
if unicode.IsControl(r) {
return ""
}
}
return v
}
// BaseSecurityHeaders returns headers that every request must carry.
func BaseSecurityHeaders() http.Header {
h := make(http.Header)
@@ -75,7 +52,7 @@ func BaseSecurityHeaders() http.Header {
h.Set(HeaderVersion, build.Version)
h.Set(HeaderBuild, DetectBuildKind())
h.Set(HeaderUserAgent, UserAgentValue())
if v := AgentTraceValue(); v != "" {
if v := envvars.AgentTrace(); v != "" {
h.Set(HeaderAgentTrace, v)
}
return h

View File

@@ -6,7 +6,6 @@ package cmdutil
import (
"context"
"net/http"
"strings"
"testing"
"github.com/larksuite/cli/extension/credential"
@@ -264,88 +263,9 @@ func TestBaseSecurityHeaders_AllRequiredHeaders(t *testing.T) {
}
// ---------------------------------------------------------------------------
// AgentTraceValue / HeaderAgentTrace
// HeaderAgentTrace injection (via BaseSecurityHeaders)
// ---------------------------------------------------------------------------
func TestAgentTraceValue_EmptyWhenEnvUnset(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, "")
if got := AgentTraceValue(); got != "" {
t.Fatalf("AgentTraceValue() = %q, want empty when env unset", got)
}
}
func TestAgentTraceValue_ReturnsCleanValue(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, "trace-abc-123")
if got := AgentTraceValue(); got != "trace-abc-123" {
t.Fatalf("AgentTraceValue() = %q, want %q", got, "trace-abc-123")
}
}
func TestAgentTraceValue_TrimsWhitespace(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, " trace-trim ")
if got := AgentTraceValue(); got != "trace-trim" {
t.Fatalf("AgentTraceValue() = %q, want %q (whitespace trimmed)", got, "trace-trim")
}
}
func TestAgentTraceValue_OnlyWhitespace_ReturnsEmpty(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, " ")
if got := AgentTraceValue(); got != "" {
t.Fatalf("AgentTraceValue() = %q, want empty for whitespace-only value", got)
}
}
func TestAgentTraceValue_RejectsCRLF(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, "val\r\nX-Evil: attack")
if got := AgentTraceValue(); got != "" {
t.Fatalf("AgentTraceValue() = %q, want empty for CR/LF value", got)
}
}
func TestAgentTraceValue_RejectsLF(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, "val\nX-Evil: attack")
if got := AgentTraceValue(); got != "" {
t.Fatalf("AgentTraceValue() = %q, want empty for LF value", got)
}
}
func TestAgentTraceValue_RejectsTab(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, "val\tinjected")
if got := AgentTraceValue(); got != "" {
t.Fatalf("AgentTraceValue() = %q, want empty for tab value", got)
}
}
func TestAgentTraceValue_RejectsControlChar(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, "val\x01injected")
if got := AgentTraceValue(); got != "" {
t.Fatalf("AgentTraceValue() = %q, want empty for control char value", got)
}
}
func TestAgentTraceValue_RejectsDEL(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, "val\x7finjected")
if got := AgentTraceValue(); got != "" {
t.Fatalf("AgentTraceValue() = %q, want empty for DEL value", got)
}
}
func TestAgentTraceValue_RejectsOverlongValue(t *testing.T) {
longVal := strings.Repeat("a", agentTraceMaxLen+1)
t.Setenv(envvars.CliAgentTrace, longVal)
if got := AgentTraceValue(); got != "" {
t.Fatalf("AgentTraceValue() returned non-empty for %d-byte value (max %d)", len(longVal), agentTraceMaxLen)
}
}
func TestAgentTraceValue_AcceptsMaxLengthValue(t *testing.T) {
val := strings.Repeat("a", agentTraceMaxLen)
t.Setenv(envvars.CliAgentTrace, val)
if got := AgentTraceValue(); got != val {
t.Fatalf("AgentTraceValue() = %q, want %d-byte value accepted", got, agentTraceMaxLen)
}
}
func TestBaseSecurityHeaders_NoAgentTraceHeaderWhenEnvUnset(t *testing.T) {
t.Setenv(envvars.CliAgentTrace, "")
h := BaseSecurityHeaders()

View File

@@ -19,6 +19,7 @@ const (
// Content safety scanning mode
CliContentSafetyMode = "LARKSUITE_CLI_CONTENT_SAFETY_MODE"
CliAgentName = "LARKSUITE_CLI_AGENT_NAME"
CliAgentTrace = "LARKSUITE_CLI_AGENT_TRACE"
CliProxyEnable = "LARKSUITE_CLI_PROXY_ENABLE"

36
internal/envvars/read.go Normal file
View File

@@ -0,0 +1,36 @@
// Copyright (c) 2026 Lark Technologies Pte. Ltd.
// SPDX-License-Identifier: MIT
package envvars
import (
"os"
"strings"
"unicode"
)
const (
agentNameMaxLen = 128
agentTraceMaxLen = 1024
)
func AgentName() string {
return sanitizeSingleLine(os.Getenv(CliAgentName), agentNameMaxLen)
}
func AgentTrace() string {
return sanitizeSingleLine(os.Getenv(CliAgentTrace), agentTraceMaxLen)
}
func sanitizeSingleLine(raw string, maxLen int) string {
v := strings.TrimSpace(raw)
if v == "" || len(v) > maxLen {
return ""
}
for _, r := range v {
if unicode.IsControl(r) {
return ""
}
}
return v
}

View File

@@ -0,0 +1,131 @@
// Copyright (c) 2026 Lark Technologies Pte. Ltd.
// SPDX-License-Identifier: MIT
package envvars
import (
"strings"
"testing"
)
func TestAgentName_EmptyWhenEnvUnset(t *testing.T) {
t.Setenv(CliAgentName, "")
if got := AgentName(); got != "" {
t.Fatalf("AgentName() = %q, want empty when env unset", got)
}
}
func TestAgentName_ReturnsCleanValue(t *testing.T) {
t.Setenv(CliAgentName, "claude-code")
if got := AgentName(); got != "claude-code" {
t.Fatalf("AgentName() = %q, want %q", got, "claude-code")
}
}
func TestAgentName_TrimsWhitespace(t *testing.T) {
t.Setenv(CliAgentName, " cursor ")
if got := AgentName(); got != "cursor" {
t.Fatalf("AgentName() = %q, want %q (whitespace trimmed)", got, "cursor")
}
}
func TestAgentName_RejectsCRLFInjection(t *testing.T) {
t.Setenv(CliAgentName, "agent\r\nX-Evil: attack")
if got := AgentName(); got != "" {
t.Fatalf("AgentName() = %q, want empty for CR/LF value", got)
}
}
func TestAgentName_RejectsControlChar(t *testing.T) {
t.Setenv(CliAgentName, "agent\x01injected")
if got := AgentName(); got != "" {
t.Fatalf("AgentName() = %q, want empty for control char value", got)
}
}
func TestAgentName_RejectsOverlongValue(t *testing.T) {
longVal := strings.Repeat("a", agentNameMaxLen+1)
t.Setenv(CliAgentName, longVal)
if got := AgentName(); got != "" {
t.Fatalf("AgentName() returned non-empty for %d-byte value (max %d)", len(longVal), agentNameMaxLen)
}
}
func TestAgentTrace_EmptyWhenEnvUnset(t *testing.T) {
t.Setenv(CliAgentTrace, "")
if got := AgentTrace(); got != "" {
t.Fatalf("AgentTrace() = %q, want empty when env unset", got)
}
}
func TestAgentTrace_ReturnsCleanValue(t *testing.T) {
t.Setenv(CliAgentTrace, "trace-abc-123")
if got := AgentTrace(); got != "trace-abc-123" {
t.Fatalf("AgentTrace() = %q, want %q", got, "trace-abc-123")
}
}
func TestAgentTrace_TrimsWhitespace(t *testing.T) {
t.Setenv(CliAgentTrace, " trace-trim ")
if got := AgentTrace(); got != "trace-trim" {
t.Fatalf("AgentTrace() = %q, want %q (whitespace trimmed)", got, "trace-trim")
}
}
func TestAgentTrace_OnlyWhitespace_ReturnsEmpty(t *testing.T) {
t.Setenv(CliAgentTrace, " ")
if got := AgentTrace(); got != "" {
t.Fatalf("AgentTrace() = %q, want empty for whitespace-only value", got)
}
}
func TestAgentTrace_RejectsCRLF(t *testing.T) {
t.Setenv(CliAgentTrace, "val\r\nX-Evil: attack")
if got := AgentTrace(); got != "" {
t.Fatalf("AgentTrace() = %q, want empty for CR/LF value", got)
}
}
func TestAgentTrace_RejectsLF(t *testing.T) {
t.Setenv(CliAgentTrace, "val\nX-Evil: attack")
if got := AgentTrace(); got != "" {
t.Fatalf("AgentTrace() = %q, want empty for LF value", got)
}
}
func TestAgentTrace_RejectsTab(t *testing.T) {
t.Setenv(CliAgentTrace, "val\tinjected")
if got := AgentTrace(); got != "" {
t.Fatalf("AgentTrace() = %q, want empty for tab value", got)
}
}
func TestAgentTrace_RejectsControlChar(t *testing.T) {
t.Setenv(CliAgentTrace, "val\x01injected")
if got := AgentTrace(); got != "" {
t.Fatalf("AgentTrace() = %q, want empty for control char value", got)
}
}
func TestAgentTrace_RejectsDEL(t *testing.T) {
t.Setenv(CliAgentTrace, "val\x7finjected")
if got := AgentTrace(); got != "" {
t.Fatalf("AgentTrace() = %q, want empty for DEL value", got)
}
}
func TestAgentTrace_RejectsOverlongValue(t *testing.T) {
longVal := strings.Repeat("a", agentTraceMaxLen+1)
t.Setenv(CliAgentTrace, longVal)
if got := AgentTrace(); got != "" {
t.Fatalf("AgentTrace() returned non-empty for %d-byte value (max %d)", len(longVal), agentTraceMaxLen)
}
}
func TestAgentTrace_AcceptsMaxLengthValue(t *testing.T) {
val := strings.Repeat("a", agentTraceMaxLen)
t.Setenv(CliAgentTrace, val)
if got := AgentTrace(); got != val {
t.Fatalf("AgentTrace() = %q, want %d-byte value accepted", got, agentTraceMaxLen)
}
}

View File

@@ -6,8 +6,7 @@ package registry
import "github.com/larksuite/cli/internal/apicatalog"
// EmbeddedCatalog returns a navigation catalog over the embedded (overlay-free)
// metadata — deterministic across machines, for `lark-cli schema`, golden tests
// and schema lint.
// metadata — deterministic across machines, for golden tests and schema lint.
func EmbeddedCatalog() apicatalog.Catalog {
return apicatalog.New(apicatalog.SourceEmbedded, EmbeddedServicesTyped())
}
@@ -18,3 +17,14 @@ func EmbeddedCatalog() apicatalog.Catalog {
func RuntimeCatalog() apicatalog.Catalog {
return apicatalog.New(apicatalog.SourceRuntime, ServicesTyped())
}
// SchemaCatalog returns the embedded catalog when metadata is compiled in,
// otherwise the merged runtime catalog. Binaries built from the bare Go module
// embed only the empty meta_data_default.json stub, so the embedded view has
// nothing to resolve; the merged view is the only data such binaries have.
func SchemaCatalog() apicatalog.Catalog {
if len(EmbeddedServicesTyped()) > 0 {
return EmbeddedCatalog()
}
return RuntimeCatalog()
}

View File

@@ -0,0 +1,67 @@
// Copyright (c) 2026 Lark Technologies Pte. Ltd.
// SPDX-License-Identifier: MIT
package registry
import (
"net/http"
"net/http/httptest"
"testing"
"github.com/larksuite/cli/internal/apicatalog"
)
// swapEmbeddedMeta replaces the compiled-in metadata bytes for one test and
// restores them (with a full state reset) on cleanup.
func swapEmbeddedMeta(t *testing.T, data []byte) {
t.Helper()
resetInit()
orig := embeddedMetaJSON
embeddedMetaJSON = data
t.Cleanup(func() {
waitBackgroundRefresh()
embeddedMetaJSON = orig
resetInit()
})
}
func TestSchemaCatalog_EmbeddedWhenCompiledIn(t *testing.T) {
swapEmbeddedMeta(t, testCacheJSON("embedded_svc"))
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", t.TempDir())
t.Setenv("LARKSUITE_CLI_REMOTE_META", "off")
c := SchemaCatalog()
if c.Source() != apicatalog.SourceEmbedded {
t.Fatalf("Source = %q, want %q", c.Source(), apicatalog.SourceEmbedded)
}
if _, ok := c.Service("embedded_svc"); !ok {
t.Fatal("expected embedded_svc from embedded metadata")
}
}
// TestSchemaCatalog_FallsBackToRuntimeWhenNoEmbedded simulates a binary built
// from the bare Go module (plugin builds): only the empty meta_data_default.json
// stub is compiled in, so SchemaCatalog must serve the merged runtime view that
// Init seeds via sync fetch.
func TestSchemaCatalog_FallsBackToRuntimeWhenNoEmbedded(t *testing.T) {
swapEmbeddedMeta(t, embeddedMetaDataDefaultJSON)
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", t.TempDir())
t.Setenv("LARKSUITE_CLI_REMOTE_META", "on")
ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
w.WriteHeader(200)
w.Write(testEnvelopeJSON("remote_svc"))
}))
defer ts.Close()
testMetaURL = ts.URL
c := SchemaCatalog()
if c.Source() != apicatalog.SourceRuntime {
t.Fatalf("Source = %q, want %q", c.Source(), apicatalog.SourceRuntime)
}
if _, ok := c.Service("remote_svc"); !ok {
t.Fatal("expected remote_svc from runtime fallback")
}
}

View File

@@ -15,6 +15,7 @@ import (
"github.com/larksuite/cli/internal/core"
"github.com/larksuite/cli/internal/meta"
"github.com/larksuite/cli/internal/update"
)
//go:embed scope_priorities.json scope_overrides.json
@@ -85,7 +86,9 @@ func InitWithBrand(brand core.LarkBrand) {
brandChanged := metaErr == nil && cm.Brand != "" && cm.Brand != string(brand)
if !brandChanged {
if cached, err := loadCachedMerged(); err == nil {
// After a CLI upgrade the embedded data can be fresher than an old
// cache; an equal/older cache must not shadow it.
if cached, err := loadCachedMerged(); err == nil && update.IsNewer(cached.Version, embeddedVersion) {
overlayMergedServices(cached)
}
}

View File

@@ -0,0 +1,102 @@
// Copyright (c) 2026 Lark Technologies Pte. Ltd.
// SPDX-License-Identifier: MIT
package registry
import (
"encoding/json"
"os"
"path/filepath"
"testing"
"time"
"github.com/larksuite/cli/internal/core"
"github.com/larksuite/cli/internal/meta"
)
// seedCache writes a cache file + cache meta for one service whose Title is
// marker, tagged with the given top-level data version and brand.
func seedCache(t *testing.T, dir, name, marker, version, brand string) {
t.Helper()
cDir := filepath.Join(dir, "cache")
if err := os.MkdirAll(cDir, 0700); err != nil {
t.Fatal(err)
}
reg := MergedRegistry{
Version: version,
Services: []meta.Service{{Name: name, Version: "cache", Title: marker}},
}
data, _ := json.Marshal(reg)
if err := os.WriteFile(filepath.Join(cDir, "remote_meta.json"), data, 0644); err != nil {
t.Fatal(err)
}
cm := CacheMeta{LastCheckAt: time.Now().Unix(), Version: version, Brand: brand}
mData, _ := json.Marshal(cm)
if err := os.WriteFile(filepath.Join(cDir, "remote_meta.meta.json"), mData, 0644); err != nil {
t.Fatal(err)
}
}
// initWithCache runs a fresh feishu-brand init with remote on, a high TTL and a
// recent LastCheckAt (so no refresh fires), embedded meta at embeddedVer and a
// pre-seeded cache at cacheVer — the overlay version gate is the only variable.
func initWithCache(t *testing.T, embeddedVer, cacheVer string) {
t.Helper()
embedded, _ := json.Marshal(MergedRegistry{
Version: embeddedVer,
Services: []meta.Service{{Name: "svc", Version: "embedded", Title: "EMBEDDED"}},
})
swapEmbeddedMeta(t, embedded)
tmp := t.TempDir()
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", tmp)
t.Setenv("LARKSUITE_CLI_REMOTE_META", "on")
t.Setenv("LARKSUITE_CLI_META_TTL", "3600")
seedCache(t, tmp, "svc", "CACHE", cacheVer, "feishu")
InitWithBrand(core.BrandFeishu)
}
func titleOf(t *testing.T, name string) string {
t.Helper()
svc, ok := ServiceTyped(name)
if !ok {
t.Fatalf("service %q not loaded", name)
}
return svc.Title
}
func TestOverlayGate_EqualVersion_UsesEmbedded(t *testing.T) {
initWithCache(t, "1.0.0", "1.0.0")
if got := titleOf(t, "svc"); got != "EMBEDDED" {
t.Errorf("equal version: got %q, want EMBEDDED (cache must not overlay)", got)
}
}
func TestOverlayGate_OlderCache_UsesEmbedded(t *testing.T) {
initWithCache(t, "2.0.0", "1.0.0")
if got := titleOf(t, "svc"); got != "EMBEDDED" {
t.Errorf("older cache: got %q, want EMBEDDED", got)
}
}
func TestOverlayGate_NewerCache_OverlaysCache(t *testing.T) {
initWithCache(t, "1.0.0", "2.0.0")
if got := titleOf(t, "svc"); got != "CACHE" {
t.Errorf("newer cache: got %q, want CACHE", got)
}
}
func TestOverlayGate_UnparseableCacheVersion_UsesEmbedded(t *testing.T) {
initWithCache(t, "1.0.0", "not-a-semver")
if got := titleOf(t, "svc"); got != "EMBEDDED" {
t.Errorf("unparseable cache version: got %q, want EMBEDDED", got)
}
}
func TestOverlayGate_StubEmbedded_OverlaysRealCache(t *testing.T) {
// The bare-module stub baseline is "0.0.0"; a real cache version must win so
// plugin builds without compiled meta_data.json still get remote data.
initWithCache(t, "0.0.0", "1.0.0")
if got := titleOf(t, "svc"); got != "CACHE" {
t.Errorf("stub-embedded baseline: got %q, want CACHE", got)
}
}

View File

@@ -72,9 +72,11 @@ func hasEmbeddedServices() bool {
}
// testRegistry returns a minimal MergedRegistry with one service.
// The version is a real semver newer than the embedded stub baseline ("0.0.0")
// so cache overlay passes the version gate in InitWithBrand.
func testRegistry(name string) MergedRegistry {
return MergedRegistry{
Version: "test-1.0",
Version: "1.0.0",
Services: []meta.Service{
{
Name: name,
@@ -160,7 +162,7 @@ func TestRemoteOff_SkipsRemoteLogic(t *testing.T) {
}
func TestCacheHit_WithinTTL(t *testing.T) {
resetInit()
swapEmbeddedMeta(t, nil) // overlay must depend only on the cache version, not the ambient embedded meta
tmp := t.TempDir()
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", tmp)
t.Setenv("LARKSUITE_CLI_REMOTE_META", "on")
@@ -197,7 +199,7 @@ func TestCacheHit_WithinTTL(t *testing.T) {
}
func TestNetworkError_SilentDegradation(t *testing.T) {
resetInit()
swapEmbeddedMeta(t, nil) // overlay must depend only on the cache version, not the ambient embedded meta
tmp := t.TempDir()
t.Setenv("LARKSUITE_CLI_CONFIG_DIR", tmp)
t.Setenv("LARKSUITE_CLI_REMOTE_META", "on")
@@ -371,8 +373,8 @@ func TestFetchRemoteMerged_200(t *testing.T) {
if data == nil {
t.Fatal("expected non-nil data")
}
if reg.Version != "test-1.0" {
t.Errorf("expected version test-1.0, got %s", reg.Version)
if reg.Version != "1.0.0" {
t.Errorf("expected version 1.0.0, got %s", reg.Version)
}
}

View File

@@ -32,6 +32,7 @@ type InstallMethod int
const (
InstallNpm InstallMethod = iota
InstallPnpm
InstallManual
)
@@ -53,22 +54,32 @@ var (
// DetectResult holds installation detection results.
type DetectResult struct {
Method InstallMethod
ResolvedPath string
NpmAvailable bool
Method InstallMethod
ResolvedPath string
NpmAvailable bool
PnpmAvailable bool
}
// CanAutoUpdate returns true if the CLI can update itself automatically.
func (d DetectResult) CanAutoUpdate() bool {
return d.Method == InstallNpm && d.NpmAvailable
switch d.Method {
case InstallNpm:
return d.NpmAvailable
case InstallPnpm:
return d.PnpmAvailable
}
return false
}
// ManualReason returns a human-readable explanation of why auto-update is unavailable.
func (d DetectResult) ManualReason() string {
if d.Method == InstallNpm && !d.NpmAvailable {
switch {
case d.Method == InstallNpm && !d.NpmAvailable:
return "installed via npm, but npm is not available in PATH"
case d.Method == InstallPnpm && !d.PnpmAvailable:
return "installed via pnpm, but pnpm is not available in PATH"
}
return "not installed via npm"
return "not installed via npm or pnpm"
}
// NpmResult holds the result of an npm install or skills update execution.
@@ -92,6 +103,7 @@ func (r *NpmResult) CombinedOutput() string {
type Updater struct {
DetectOverride func() DetectResult
NpmInstallOverride func(version string) *NpmResult
PnpmInstallOverride func(version string) *NpmResult
SkillsIndexFetchOverride func() *NpmResult
SkillsCommandOverride func(args ...string) *NpmResult
VerifyOverride func(expectedVersion string) error
@@ -101,17 +113,38 @@ type Updater struct {
// running binary is successfully renamed to .old. Used by
// CanRestorePreviousVersion to report whether rollback is possible.
backupCreated bool
// detectCache memoizes the first real DetectInstallMethod result. How this
// binary was installed cannot change during a single process, so caching is
// the correct semantics — and it is required for correctness: the update
// flow mutates the install (pnpm add -g / npm install -g) before syncing
// skills, so a re-detection at skills time could resolve a now-stale
// os.Executable path and misclassify. Seeded pre-update by the first call
// (updateRun), it keeps the post-update skills launcher consistent with the
// launcher reported to the user. Not goroutine-safe; the update flow is
// sequential.
detectCache *DetectResult
}
// New creates an Updater with default (real) behavior.
func New() *Updater { return &Updater{} }
// DetectInstallMethod determines how the CLI was installed and whether
// npm is available for auto-update.
// DetectInstallMethod determines how the CLI was installed and whether the
// owning package manager is available for auto-update.
func (u *Updater) DetectInstallMethod() DetectResult {
if u.DetectOverride != nil {
return u.DetectOverride()
}
if u.detectCache != nil {
return *u.detectCache
}
result := u.detectInstallMethod()
u.detectCache = &result
return result
}
// detectInstallMethod performs the real (uncached) detection.
func (u *Updater) detectInstallMethod() DetectResult {
exe, err := vfs.Executable()
if err != nil {
return DetectResult{Method: InstallManual}
@@ -120,24 +153,54 @@ func (u *Updater) DetectInstallMethod() DetectResult {
if err != nil {
return DetectResult{Method: InstallManual, ResolvedPath: exe}
}
_, npmErr := exec.LookPath("npm")
_, pnpmErr := exec.LookPath("pnpm")
return detectFromResolved(resolved, npmErr == nil, pnpmErr == nil)
}
// detectFromResolved classifies the resolved binary path into an install
// method and records package-manager availability. Split out from
// DetectInstallMethod so the classification is unit-testable without touching
// the filesystem or PATH.
func detectFromResolved(resolved string, npmOnPath, pnpmOnPath bool) DetectResult {
method := InstallManual
if strings.Contains(resolved, "node_modules") {
method = InstallNpm
}
npmAvailable := false
if method == InstallNpm {
if _, err := exec.LookPath("npm"); err == nil {
npmAvailable = true
if containsPnpmMarker(resolved) {
method = InstallPnpm
} else {
method = InstallNpm
}
}
return DetectResult{
Method: method,
ResolvedPath: resolved,
NpmAvailable: npmAvailable,
d := DetectResult{Method: method, ResolvedPath: resolved}
switch method {
case InstallNpm:
d.NpmAvailable = npmOnPath
case InstallPnpm:
d.PnpmAvailable = pnpmOnPath
}
return d
}
// containsPnpmMarker reports whether the resolved binary path belongs to a
// pnpm-managed install. pnpm exposes two layouts: the classic virtual store
// (a ".pnpm" directory segment) and the global content-addressable store,
// whose resolved path runs through pnpm's home directory (e.g.
// "~/Library/pnpm/store/v11/links/...") — a "pnpm" segment immediately
// followed by "store". Matching only these two shapes (rather than any bare
// "pnpm" segment) avoids misclassifying an npm install that merely lives under
// a directory named "pnpm". Windows separators are normalized to "/" so the
// classification is OS-independent and unit-testable anywhere.
func containsPnpmMarker(p string) bool {
parts := strings.Split(strings.ReplaceAll(p, `\`, "/"), "/")
for i, part := range parts {
if part == ".pnpm" {
return true
}
if part == "pnpm" && i+1 < len(parts) && parts[i+1] == "store" {
return true
}
}
return false
}
// RunNpmInstall executes npm install -g @larksuite/cli@<version>.
@@ -163,6 +226,29 @@ func (u *Updater) RunNpmInstall(version string) *NpmResult {
return r
}
// RunPnpmInstall executes pnpm add -g @larksuite/cli@<version>.
func (u *Updater) RunPnpmInstall(version string) *NpmResult {
if u.PnpmInstallOverride != nil {
return u.PnpmInstallOverride(version)
}
r := &NpmResult{}
pnpmPath, err := exec.LookPath("pnpm")
if err != nil {
r.Err = fmt.Errorf("pnpm not found in PATH: %w", err)
return r
}
ctx, cancel := context.WithTimeout(context.Background(), npmInstallTimeout)
defer cancel()
cmd := exec.CommandContext(ctx, pnpmPath, "add", "-g", NpmPackage+"@"+version)
cmd.Stdout = &r.Stdout
cmd.Stderr = &r.Stderr
r.Err = cmd.Run()
if ctx.Err() == context.DeadlineExceeded {
r.Err = fmt.Errorf("pnpm install timed out after %s", npmInstallTimeout)
}
return r
}
func (u *Updater) ListOfficialSkillsIndex() *NpmResult {
if u.SkillsIndexFetchOverride != nil {
return u.SkillsIndexFetchOverride()
@@ -261,19 +347,40 @@ func (u *Updater) runSkillsInstall(source string, nameList []string) *NpmResult
return u.runSkillsCommand(args...)
}
// skillsInvocation decides how to launch the `skills` CLI. When the lark-cli
// itself was installed via pnpm and pnpm is available, it uses `pnpm dlx` so
// pnpm-only environments (pnpm's standalone installer bundles Node without
// putting npm/npx on PATH) can still sync skills after a self-update.
// Otherwise it uses `npx`. The npx auto-confirm flag "-y", when present as the
// leading arg, maps to `pnpm dlx`'s default non-interactive behavior and is
// dropped for the pnpm launcher. Kept pure (no exec/PATH access) so the
// launcher selection is unit-testable on any platform.
func skillsInvocation(method InstallMethod, pnpmAvailable bool, args []string) (launcher string, rest []string) {
if method == InstallPnpm && pnpmAvailable {
r := args
if len(r) > 0 && r[0] == "-y" {
r = r[1:]
}
return "pnpm", append([]string{"dlx"}, r...)
}
return "npx", args
}
func (u *Updater) runSkillsCommand(args ...string) *NpmResult {
if u.SkillsCommandOverride != nil {
return u.SkillsCommandOverride(args...)
}
r := &NpmResult{}
npxPath, err := exec.LookPath("npx")
det := u.DetectInstallMethod()
launcher, cmdArgs := skillsInvocation(det.Method, det.PnpmAvailable, args)
binPath, err := exec.LookPath(launcher)
if err != nil {
r.Err = fmt.Errorf("npx not found in PATH: %w", err)
r.Err = fmt.Errorf("%s not found in PATH: %w", launcher, err)
return r
}
ctx, cancel := context.WithTimeout(context.Background(), skillsUpdateTimeout)
defer cancel()
cmd := exec.CommandContext(ctx, npxPath, args...)
cmd := exec.CommandContext(ctx, binPath, cmdArgs...)
cmd.Stdout = &r.Stdout
cmd.Stderr = &r.Stderr
r.Err = cmd.Run()

View File

@@ -371,3 +371,147 @@ func TestListOfficialSkillsFallsBack(t *testing.T) {
t.Fatalf("fallback call = %q, want larksuite/cli --list", called[1])
}
}
func TestContainsPnpmMarker(t *testing.T) {
cases := []struct {
path string
want bool
}{
// Classic virtual-store layout (.pnpm segment).
{"/Users/x/Library/pnpm/global/5/node_modules/.pnpm/@larksuite+cli@1.0.44/node_modules/@larksuite/cli/bin/lark-cli", true},
{`C:\Users\x\AppData\Local\pnpm\global\5\node_modules\.pnpm\@larksuite+cli@1.0.44\node_modules\@larksuite\cli\bin\lark-cli.exe`, true},
// Global content-addressable store layout (pnpm 11): resolved path runs
// through the pnpm home store, a "pnpm" segment with no ".pnpm".
{"/Users/x/Library/pnpm/store/v11/links/@larksuite/cli/1.0.59/abc123/node_modules/@larksuite/cli/bin/lark-cli", true},
{"/home/x/.local/share/pnpm/store/v10/@larksuite/cli/node_modules/@larksuite/cli/bin/lark-cli", true},
{`C:\Users\x\AppData\Local\pnpm\store\v11\links\@larksuite\cli\node_modules\@larksuite\cli\bin\lark-cli.exe`, true},
// npm and non-package installs — no pnpm/.pnpm segment.
{"/usr/local/lib/node_modules/@larksuite/cli/bin/lark-cli", false},
{"/usr/local/bin/lark-cli", false},
// Substrings that must NOT match: segment must be exactly .pnpm, or
// "pnpm" immediately followed by "store".
{"/opt/homebrew/.pnpmfoo/node_modules/@larksuite/cli/bin/lark-cli", false},
{"/opt/pnpmfoo/node_modules/@larksuite/cli/bin/lark-cli", false},
// A bare "pnpm" directory NOT followed by "store" (e.g. an npm install
// living under a dir named pnpm) must not be misclassified as pnpm.
{"/opt/pnpm/lib/node_modules/@larksuite/cli/bin/lark-cli", false},
}
for _, c := range cases {
if got := containsPnpmMarker(c.path); got != c.want {
t.Errorf("containsPnpmMarker(%q) = %v, want %v", c.path, got, c.want)
}
}
}
func TestDetectInstallMethod_Pnpm(t *testing.T) {
u := &Updater{DetectOverride: nil}
u.DetectOverride = func() DetectResult {
// Exercise the real classification by feeding a resolved path via a small shim.
return detectFromResolved("/x/node_modules/.pnpm/@larksuite+cli@1.0.44/node_modules/@larksuite/cli/bin/lark-cli", true, true)
}
got := u.DetectInstallMethod()
if got.Method != InstallPnpm {
t.Errorf("Method = %v, want InstallPnpm", got.Method)
}
if !got.PnpmAvailable {
t.Errorf("PnpmAvailable = false, want true")
}
}
func TestDetectInstallMethod_NpmVsManual(t *testing.T) {
if m := detectFromResolved("/usr/local/lib/node_modules/@larksuite/cli/bin/lark-cli", true, false).Method; m != InstallNpm {
t.Errorf("npm path Method = %v, want InstallNpm", m)
}
if m := detectFromResolved("/usr/local/bin/lark-cli", false, false).Method; m != InstallManual {
t.Errorf("manual path Method = %v, want InstallManual", m)
}
}
func TestCanAutoUpdate_Pnpm(t *testing.T) {
if !(DetectResult{Method: InstallPnpm, PnpmAvailable: true}).CanAutoUpdate() {
t.Error("pnpm available should CanAutoUpdate")
}
if (DetectResult{Method: InstallPnpm, PnpmAvailable: false}).CanAutoUpdate() {
t.Error("pnpm unavailable should not CanAutoUpdate")
}
}
func TestManualReason_Pnpm(t *testing.T) {
if got := (DetectResult{Method: InstallPnpm, NpmAvailable: false, PnpmAvailable: false}).ManualReason(); got != "installed via pnpm, but pnpm is not available in PATH" {
t.Errorf("pnpm reason = %q", got)
}
if got := (DetectResult{Method: InstallManual}).ManualReason(); got != "not installed via npm or pnpm" {
t.Errorf("manual reason = %q", got)
}
}
func TestRunPnpmInstall_Override(t *testing.T) {
u := &Updater{PnpmInstallOverride: func(version string) *NpmResult {
r := &NpmResult{}
r.Stdout.WriteString("added @larksuite/cli@" + version)
return r
}}
got := u.RunPnpmInstall("2.0.0")
if got.Err != nil {
t.Fatalf("unexpected err: %v", got.Err)
}
if !strings.Contains(got.CombinedOutput(), "2.0.0") {
t.Errorf("output = %q, want version echoed", got.CombinedOutput())
}
}
func TestRunPnpmInstall_Error(t *testing.T) {
wantErr := errors.New("boom")
u := &Updater{PnpmInstallOverride: func(string) *NpmResult { return &NpmResult{Err: wantErr} }}
if got := u.RunPnpmInstall("2.0.0"); !errors.Is(got.Err, wantErr) {
t.Errorf("err = %v, want %v", got.Err, wantErr)
}
}
func TestSkillsInvocation(t *testing.T) {
addArgs := []string{"-y", "skills", "add", "https://open.feishu.cn", "-g", "-y"}
cases := []struct {
name string
method InstallMethod
pnpmAvailable bool
args []string
wantLauncher string
wantRest []string
}{
{"pnpm install + pnpm available → pnpm dlx, drop leading -y", InstallPnpm, true, addArgs,
"pnpm", []string{"dlx", "skills", "add", "https://open.feishu.cn", "-g", "-y"}},
{"pnpm install but pnpm unavailable → npx unchanged", InstallPnpm, false, addArgs,
"npx", addArgs},
{"npm install → npx unchanged", InstallNpm, false, addArgs,
"npx", addArgs},
{"manual install → npx unchanged", InstallManual, false, []string{"-y", "skills", "ls", "-g"},
"npx", []string{"-y", "skills", "ls", "-g"}},
{"pnpm without a leading -y → prepend dlx only", InstallPnpm, true, []string{"skills", "ls", "-g"},
"pnpm", []string{"dlx", "skills", "ls", "-g"}},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
gotLauncher, gotRest := skillsInvocation(c.method, c.pnpmAvailable, c.args)
if gotLauncher != c.wantLauncher {
t.Errorf("launcher = %q, want %q", gotLauncher, c.wantLauncher)
}
if strings.Join(gotRest, " ") != strings.Join(c.wantRest, " ") {
t.Errorf("rest = %v, want %v", gotRest, c.wantRest)
}
})
}
}
// TestDetectInstallMethod_Caches locks the fix for the post-update re-detection
// hazard: DetectInstallMethod must return the first (pre-update) detection on
// subsequent calls, so the skills launcher chosen after the binary is replaced
// stays consistent with what was detected — and reported — before the update.
func TestDetectInstallMethod_Caches(t *testing.T) {
u := New()
cached := DetectResult{Method: InstallPnpm, PnpmAvailable: true, ResolvedPath: "/x/pnpm/store/v11/links/@larksuite/cli/1.0.0/node_modules/@larksuite/cli/bin/lark-cli"}
u.detectCache = &cached
got := u.DetectInstallMethod()
if got.Method != InstallPnpm || !got.PnpmAvailable {
t.Errorf("expected cached pnpm result to be returned, got %+v", got)
}
}

View File

@@ -1,6 +1,6 @@
{
"name": "@larksuite/cli",
"version": "1.0.64",
"version": "1.0.65",
"description": "The official CLI for Lark/Feishu open platform",
"bin": {
"lark-cli": "scripts/run.js"

View File

@@ -67,6 +67,26 @@ func parseAttendees(attendeesStr string, currentUserId string) ([]map[string]str
return attendees, nil
}
func attendeesIncludeRoom(attendees []map[string]string) bool {
for _, attendee := range attendees {
if attendee["type"] == "resource" || attendee["room_id"] != "" {
return true
}
}
return false
}
func guideApprovalRoomReasonError(err error, attendees []map[string]string) error {
if err == nil || !attendeesIncludeRoom(attendees) {
return err
}
p, ok := errs.ProblemOf(err)
if !ok || !strings.Contains(strings.ToLower(p.Hint), "approval_reason") {
return err
}
return withStepContext(err, "approval meeting rooms require attendees[].approval_reason; calendar +create does not expose this low-frequency field. Create the event with the raw API flow, then use `lark-cli calendar event.attendees create --as user` with attendees[].approval_reason for the room attendee.")
}
var CalendarCreate = common.Shortcut{
Service: "calendar",
Command: "+create",
@@ -225,6 +245,7 @@ var CalendarCreate = common.Shortcut{
"need_notification": true,
})
if err != nil {
err = guideApprovalRoomReasonError(err, attendees)
// Rollback: delete the event
_, rollbackErr := runtime.CallAPITyped("DELETE",
fmt.Sprintf("/open-apis/calendar/v4/calendars/%s/events/%s", validate.EncodePathSegment(calendarId), validate.EncodePathSegment(eventId)),

View File

@@ -673,6 +673,76 @@ func TestCreate_WithAttendees_InvalidParamsWithDetail_RollsBack(t *testing.T) {
}
}
func TestCreate_ApprovalRoomMissingReason_GuidesRawAttendeesAPI(t *testing.T) {
f, _, _, reg := cmdutil.TestFactory(t, defaultConfig())
reg.Register(&httpmock.Stub{
Method: "POST",
URL: "/open-apis/calendar/v4/calendars/cal_test123/events",
Body: map[string]interface{}{
"code": 0, "msg": "ok",
"data": map[string]interface{}{
"event": map[string]interface{}{
"event_id": "evt_approval_room",
"summary": "Approval Room",
"start_time": map[string]interface{}{"timestamp": "1742515200"},
"end_time": map[string]interface{}{"timestamp": "1742518800"},
},
},
},
})
reg.Register(&httpmock.Stub{
Method: "POST",
URL: "/events/evt_approval_room/attendees",
Body: map[string]interface{}{
"code": codeInvalidParamsWithDetail,
"msg": "invalid params",
"error": map[string]interface{}{
"details": []interface{}{
map[string]interface{}{"value": "attendees[0].approval_reason is required for approval meeting rooms"},
},
},
},
})
reg.Register(&httpmock.Stub{
Method: "DELETE",
URL: "/events/evt_approval_room",
Body: map[string]interface{}{"code": 0, "msg": "ok"},
})
err := mountAndRun(t, CalendarCreate, []string{
"+create",
"--summary", "Approval Room",
"--start", "2025-03-21T00:00:00+08:00",
"--end", "2025-03-21T01:00:00+08:00",
"--calendar-id", "cal_test123",
"--attendee-ids", "omm_room1",
"--as", "user",
}, f, nil)
if err == nil {
t.Fatal("expected error for approval room missing approval_reason, got nil")
}
p, ok := errs.ProblemOf(err)
if !ok {
t.Fatalf("ProblemOf returned !ok for %T", err)
}
if p.Category != errs.CategoryAPI {
t.Errorf("category=%q, want %q", p.Category, errs.CategoryAPI)
}
if p.Subtype != errs.SubtypeInvalidParameters {
t.Errorf("subtype=%q, want %q", p.Subtype, errs.SubtypeInvalidParameters)
}
if p.Code != codeInvalidParamsWithDetail {
t.Errorf("code=%d, want %d", p.Code, codeInvalidParamsWithDetail)
}
for _, want := range []string{"approval_reason", "calendar event.attendees create", "--as user", "rolled back successfully"} {
if !strings.Contains(p.Hint, want) {
t.Errorf("hint should contain %q, got: %q", want, p.Hint)
}
}
}
// When the add-attendees call fails AND the rollback DELETE also fails, the
// primary error stays the add failure (classification preserved) and the Hint
// must surface BOTH the rollback failure reason and the orphan event_id so the

View File

@@ -0,0 +1,261 @@
// Copyright (c) 2026 Lark Technologies Pte. Ltd.
// SPDX-License-Identifier: MIT
package doc
import (
"context"
"fmt"
"net/http"
"strconv"
"strings"
"github.com/larksuite/cli/errs"
"github.com/larksuite/cli/internal/validate"
"github.com/larksuite/cli/shortcuts/common"
)
type docsHistoryListSpec struct {
Doc documentRef
PageSize int
PageToken string
}
type docsHistoryRevertSpec struct {
Doc documentRef
HistoryVersionID string
WaitTimeoutMs int
}
type docsHistoryRevertStatusSpec struct {
Doc documentRef
TaskID string
}
func parseDocsHistoryDocRef(raw, shortcut string) (documentRef, error) {
ref, err := parseDocumentRef(raw)
if err != nil {
return documentRef{}, err
}
if ref.Kind == "doc" {
return documentRef{}, errs.NewValidationError(errs.SubtypeInvalidArgument, "docs %s only supports docx documents; use a docx token/URL or a wiki URL that resolves to docx", shortcut).WithParam("--doc")
}
return ref, nil
}
func validateDocsHistoryPageSize(pageSize int) error {
if pageSize < 1 || pageSize > 20 {
return errs.NewValidationError(errs.SubtypeInvalidArgument, "invalid --page-size %d: must be between 1 and 20", pageSize).WithParam("--page-size")
}
return nil
}
func validateDocsHistoryVersionID(historyVersionID string) error {
version, err := strconv.ParseInt(strings.TrimSpace(historyVersionID), 10, 64)
if err != nil {
return errs.NewValidationError(errs.SubtypeInvalidArgument, "--history-version-id must be a positive integer string returned by docs +history-list").WithParam("--history-version-id").WithCause(err)
}
if version <= 0 {
return errs.NewValidationError(errs.SubtypeInvalidArgument, "--history-version-id must be a positive integer string returned by docs +history-list").WithParam("--history-version-id")
}
return nil
}
func validateDocsHistoryWaitTimeout(timeoutMs int) error {
if timeoutMs < 0 || timeoutMs > 30000 {
return errs.NewValidationError(errs.SubtypeInvalidArgument, "invalid --wait-timeout-ms %d: must be between 0 and 30000", timeoutMs).WithParam("--wait-timeout-ms")
}
return nil
}
func docsHistoryListParams(spec docsHistoryListSpec) map[string]interface{} {
params := map[string]interface{}{
"page_size": spec.PageSize,
}
if spec.PageToken != "" {
params["page_token"] = spec.PageToken
}
return params
}
func docsHistoryRevertBody(spec docsHistoryRevertSpec) map[string]interface{} {
return map[string]interface{}{
"history_version_id": spec.HistoryVersionID,
"wait_timeout_ms": spec.WaitTimeoutMs,
}
}
func docsHistoryStatusParams(spec docsHistoryRevertStatusSpec) map[string]interface{} {
return map[string]interface{}{
"task_id": spec.TaskID,
}
}
func docsHistoryAPIPath(docToken, suffix string) string {
return fmt.Sprintf("/open-apis/docs_ai/v1/documents/%s/%s", validate.EncodePathSegment(docToken), suffix)
}
var DocsHistoryList = common.Shortcut{
Service: "docs",
Command: "+history-list",
Description: "List Lark document history versions",
Risk: "read",
Scopes: []string{"docx:document:readonly"},
AuthTypes: []string{"user", "bot"},
PostMount: installDocsShortcutHelp("+history-list"),
Flags: []common.Flag{
{Name: "doc", Desc: "document URL or token", Required: true},
{Name: "page-size", Type: "int", Default: "20", Desc: "history entries to return, range 1-20"},
{Name: "page-token", Desc: "pagination token from the previous page's page_token"},
},
Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
if _, err := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-list"); err != nil {
return err
}
return validateDocsHistoryPageSize(runtime.Int("page-size"))
},
DryRun: func(ctx context.Context, runtime *common.RuntimeContext) *common.DryRunAPI {
ref, _ := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-list")
spec := docsHistoryListSpec{
Doc: ref,
PageSize: runtime.Int("page-size"),
PageToken: strings.TrimSpace(runtime.Str("page-token")),
}
return common.NewDryRunAPI().
Desc("OpenAPI: list document history versions").
GET("/open-apis/docs_ai/v1/documents/:document_id/histories").
Set("document_id", spec.Doc.Token).
Params(docsHistoryListParams(spec))
},
Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
ref, _ := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-list")
spec := docsHistoryListSpec{
Doc: ref,
PageSize: runtime.Int("page-size"),
PageToken: strings.TrimSpace(runtime.Str("page-token")),
}
data, err := runtime.CallAPITyped(
http.MethodGet,
docsHistoryAPIPath(spec.Doc.Token, "histories"),
docsHistoryListParams(spec),
nil,
)
if err != nil {
return err
}
runtime.OutRaw(data, nil)
return nil
},
}
var DocsHistoryRevert = common.Shortcut{
Service: "docs",
Command: "+history-revert",
Description: "Revert a Lark document to a historical version",
Risk: "write",
Scopes: []string{"docx:document:write_only", "docx:document:readonly"},
AuthTypes: []string{"user", "bot"},
PostMount: installDocsShortcutHelp("+history-revert"),
Flags: []common.Flag{
{Name: "doc", Desc: "document URL or token", Required: true},
{Name: "history-version-id", Desc: "history_version_id from docs +history-list to revert to", Required: true},
{Name: "wait-timeout-ms", Type: "int", Default: "30000", Desc: "milliseconds to wait for revert completion before returning, range 0-30000"},
},
Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
if _, err := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-revert"); err != nil {
return err
}
if err := validateDocsHistoryVersionID(runtime.Str("history-version-id")); err != nil {
return err
}
return validateDocsHistoryWaitTimeout(runtime.Int("wait-timeout-ms"))
},
DryRun: func(ctx context.Context, runtime *common.RuntimeContext) *common.DryRunAPI {
ref, _ := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-revert")
spec := docsHistoryRevertSpec{
Doc: ref,
HistoryVersionID: strings.TrimSpace(runtime.Str("history-version-id")),
WaitTimeoutMs: runtime.Int("wait-timeout-ms"),
}
return common.NewDryRunAPI().
Desc("OpenAPI: revert document history").
POST("/open-apis/docs_ai/v1/documents/:document_id/history/revert").
Set("document_id", spec.Doc.Token).
Body(docsHistoryRevertBody(spec))
},
Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
ref, _ := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-revert")
spec := docsHistoryRevertSpec{
Doc: ref,
HistoryVersionID: strings.TrimSpace(runtime.Str("history-version-id")),
WaitTimeoutMs: runtime.Int("wait-timeout-ms"),
}
data, err := runtime.CallAPITyped(
http.MethodPost,
docsHistoryAPIPath(spec.Doc.Token, "history/revert"),
nil,
docsHistoryRevertBody(spec),
)
if err != nil {
return err
}
runtime.OutRaw(data, nil)
return nil
},
}
var DocsHistoryRevertStatus = common.Shortcut{
Service: "docs",
Command: "+history-revert-status",
Description: "Get Lark document history revert task status",
Risk: "read",
Scopes: []string{"docx:document:readonly"},
AuthTypes: []string{"user", "bot"},
PostMount: installDocsShortcutHelp("+history-revert-status"),
Flags: []common.Flag{
{Name: "doc", Desc: "document URL or token", Required: true},
{Name: "task-id", Desc: "task_id returned by docs +history-revert", Required: true},
},
Validate: func(ctx context.Context, runtime *common.RuntimeContext) error {
if _, err := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-revert-status"); err != nil {
return err
}
if strings.TrimSpace(runtime.Str("task-id")) == "" {
return errs.NewValidationError(errs.SubtypeInvalidArgument, "--task-id is required").WithParam("--task-id")
}
return nil
},
DryRun: func(ctx context.Context, runtime *common.RuntimeContext) *common.DryRunAPI {
ref, _ := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-revert-status")
spec := docsHistoryRevertStatusSpec{
Doc: ref,
TaskID: strings.TrimSpace(runtime.Str("task-id")),
}
return common.NewDryRunAPI().
Desc("OpenAPI: get document history revert status").
GET("/open-apis/docs_ai/v1/documents/:document_id/history/revert_status").
Set("document_id", spec.Doc.Token).
Params(docsHistoryStatusParams(spec))
},
Execute: func(ctx context.Context, runtime *common.RuntimeContext) error {
ref, _ := parseDocsHistoryDocRef(runtime.Str("doc"), "+history-revert-status")
spec := docsHistoryRevertStatusSpec{
Doc: ref,
TaskID: strings.TrimSpace(runtime.Str("task-id")),
}
data, err := runtime.CallAPITyped(
http.MethodGet,
docsHistoryAPIPath(spec.Doc.Token, "history/revert_status"),
docsHistoryStatusParams(spec),
nil,
)
if err != nil {
return err
}
runtime.OutRaw(data, nil)
return nil
},
}

View File

@@ -0,0 +1,340 @@
// Copyright (c) 2026 Lark Technologies Pte. Ltd.
// SPDX-License-Identifier: MIT
package doc
import (
"bytes"
"context"
"encoding/json"
"errors"
"strings"
"testing"
"github.com/larksuite/cli/errs"
"github.com/larksuite/cli/internal/cmdutil"
"github.com/larksuite/cli/internal/httpmock"
"github.com/larksuite/cli/shortcuts/common"
"github.com/spf13/cobra"
)
func TestDocsHistoryValidation(t *testing.T) {
t.Parallel()
tests := []struct {
name string
shortcut common.Shortcut
args []string
param string
category errs.Category
subtype errs.Subtype
wantCause bool
}{
{
name: "list rejects legacy doc URL",
shortcut: DocsHistoryList,
args: []string{"+history-list", "--doc", "https://example.feishu.cn/doc/old_doc", "--as", "bot"},
param: "--doc",
category: errs.CategoryValidation,
subtype: errs.SubtypeInvalidArgument,
},
{
name: "list rejects invalid page size",
shortcut: DocsHistoryList,
args: []string{"+history-list", "--doc", "doxcnHistory", "--page-size", "0", "--as", "bot"},
param: "--page-size",
category: errs.CategoryValidation,
subtype: errs.SubtypeInvalidArgument,
},
{
name: "revert rejects non-numeric history version id",
shortcut: DocsHistoryRevert,
args: []string{"+history-revert", "--doc", "doxcnHistory", "--history-version-id", "abc", "--as", "bot"},
param: "--history-version-id",
category: errs.CategoryValidation,
subtype: errs.SubtypeInvalidArgument,
wantCause: true,
},
{
name: "revert rejects non-positive history version id",
shortcut: DocsHistoryRevert,
args: []string{"+history-revert", "--doc", "doxcnHistory", "--history-version-id", "0", "--as", "bot"},
param: "--history-version-id",
category: errs.CategoryValidation,
subtype: errs.SubtypeInvalidArgument,
},
{
name: "revert rejects invalid wait timeout",
shortcut: DocsHistoryRevert,
args: []string{"+history-revert", "--doc", "doxcnHistory", "--history-version-id", "10", "--wait-timeout-ms", "30001", "--as", "bot"},
param: "--wait-timeout-ms",
category: errs.CategoryValidation,
subtype: errs.SubtypeInvalidArgument,
},
{
name: "status rejects empty task id",
shortcut: DocsHistoryRevertStatus,
args: []string{"+history-revert-status", "--doc", "doxcnHistory", "--task-id", "", "--as", "bot"},
param: "--task-id",
category: errs.CategoryValidation,
subtype: errs.SubtypeInvalidArgument,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
t.Parallel()
f, stdout, _, _ := cmdutil.TestFactory(t, docsTestConfigWithAppID("docs-history-validation"))
err := mountAndRunDocs(t, tt.shortcut, tt.args, f, stdout)
if err == nil {
t.Fatal("expected validation error, got nil")
}
problem, ok := errs.ProblemOf(err)
if !ok {
t.Fatalf("error is not typed: %T %v", err, err)
}
if problem.Category != tt.category {
t.Fatalf("category = %q, want %q (err: %v)", problem.Category, tt.category, err)
}
if problem.Subtype != tt.subtype {
t.Fatalf("subtype = %q, want %q (err: %v)", problem.Subtype, tt.subtype, err)
}
var validationErr *errs.ValidationError
if !errors.As(err, &validationErr) {
t.Fatalf("expected validation error, got %T: %v", err, err)
}
if validationErr.Param != tt.param {
t.Fatalf("param = %q, want %q (err: %v)", validationErr.Param, tt.param, err)
}
if tt.wantCause && errors.Unwrap(err) == nil {
t.Fatalf("expected wrapped cause, got nil (err: %v)", err)
}
})
}
}
func TestDocsHistoryDryRun(t *testing.T) {
t.Parallel()
listCmd := newDocsHistoryRuntimeCmd(t, DocsHistoryList, map[string]string{
"doc": "doxcnHistoryDryRun",
"page-size": "5",
"page-token": "page_token_1",
})
listDry := decodeDocDryRun(t, DocsHistoryList.DryRun(context.Background(), common.TestNewRuntimeContext(listCmd, nil)))
if got, want := listDry.API[0].URL, "/open-apis/docs_ai/v1/documents/doxcnHistoryDryRun/histories"; got != want {
t.Fatalf("list dry-run URL = %q, want %q", got, want)
}
if got := int(listDry.API[0].Params["page_size"].(float64)); got != 5 {
t.Fatalf("list page_size = %d, want 5", got)
}
if got := listDry.API[0].Params["page_token"]; got != "page_token_1" {
t.Fatalf("list page_token = %#v, want page_token_1", got)
}
revertCmd := newDocsHistoryRuntimeCmd(t, DocsHistoryRevert, map[string]string{
"doc": "doxcnHistoryDryRun",
"history-version-id": "42",
"wait-timeout-ms": "30000",
})
revertDry := decodeDocDryRun(t, DocsHistoryRevert.DryRun(context.Background(), common.TestNewRuntimeContext(revertCmd, nil)))
if got, want := revertDry.API[0].URL, "/open-apis/docs_ai/v1/documents/doxcnHistoryDryRun/history/revert"; got != want {
t.Fatalf("revert dry-run URL = %q, want %q", got, want)
}
if got := revertDry.API[0].Body["history_version_id"]; got != "42" {
t.Fatalf("revert history_version_id = %#v, want 42", got)
}
if got := int(revertDry.API[0].Body["wait_timeout_ms"].(float64)); got != 30000 {
t.Fatalf("revert wait_timeout_ms = %d, want 30000", got)
}
statusCmd := newDocsHistoryRuntimeCmd(t, DocsHistoryRevertStatus, map[string]string{
"doc": "doxcnHistoryDryRun",
"task-id": "task_1",
})
statusDry := decodeDocDryRun(t, DocsHistoryRevertStatus.DryRun(context.Background(), common.TestNewRuntimeContext(statusCmd, nil)))
if got, want := statusDry.API[0].URL, "/open-apis/docs_ai/v1/documents/doxcnHistoryDryRun/history/revert_status"; got != want {
t.Fatalf("status dry-run URL = %q, want %q", got, want)
}
if got := statusDry.API[0].Params["task_id"]; got != "task_1" {
t.Fatalf("status task_id = %#v, want task_1", got)
}
}
func TestDocsHistoryExecuteList(t *testing.T) {
f, stdout, _, reg := cmdutil.TestFactory(t, docsTestConfigWithAppID("docs-history-list"))
stub := &httpmock.Stub{
Method: "GET",
URL: "/open-apis/docs_ai/v1/documents/doxcnHistory/histories",
Body: map[string]interface{}{
"code": 0,
"msg": "ok",
"data": map[string]interface{}{
"entries": []interface{}{
map[string]interface{}{
"revision_id": float64(42),
"history_version_id": "11",
"edit_time": "1780000000",
"type": float64(1),
"editor_ids": []interface{}{"ou_1"},
},
},
"has_more": true,
"page_token": "page_token_2",
},
},
}
reg.Register(stub)
err := mountAndRunDocs(t, DocsHistoryList, []string{
"+history-list",
"--doc", "doxcnHistory",
"--page-size", "5",
"--page-token", "page_token_1",
"--as", "bot",
}, f, stdout)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
data := decodeDocsHistoryEnvelope(t, stdout)
if got := data["page_token"]; got != "page_token_2" {
t.Fatalf("page_token = %#v, want page_token_2", got)
}
entries, _ := data["entries"].([]interface{})
if len(entries) != 1 {
t.Fatalf("entries = %#v, want one entry", data["entries"])
}
}
func TestDocsHistoryExecuteRevert(t *testing.T) {
f, stdout, _, reg := cmdutil.TestFactory(t, docsTestConfigWithAppID("docs-history-revert"))
stub := &httpmock.Stub{
Method: "POST",
URL: "/open-apis/docs_ai/v1/documents/doxcnHistory/history/revert",
Body: map[string]interface{}{
"code": 0,
"msg": "ok",
"data": map[string]interface{}{
"task_id": "task_1",
"status": "running",
"history_version_id": "42",
"poll_after_ms": float64(10000),
},
},
}
reg.Register(stub)
err := mountAndRunDocs(t, DocsHistoryRevert, []string{
"+history-revert",
"--doc", "doxcnHistory",
"--history-version-id", "42",
"--wait-timeout-ms", "0",
"--as", "bot",
}, f, stdout)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
var body map[string]interface{}
if err := json.Unmarshal(stub.CapturedBody, &body); err != nil {
t.Fatalf("decode revert body: %v\nraw=%s", err, stub.CapturedBody)
}
if got := body["history_version_id"]; got != "42" {
t.Fatalf("history_version_id = %#v, want 42", got)
}
if got := int(body["wait_timeout_ms"].(float64)); got != 0 {
t.Fatalf("wait_timeout_ms = %d, want 0", got)
}
data := decodeDocsHistoryEnvelope(t, stdout)
if got := data["task_id"]; got != "task_1" {
t.Fatalf("task_id = %#v, want task_1", got)
}
}
func TestDocsHistoryExecuteRevertStatus(t *testing.T) {
f, stdout, _, reg := cmdutil.TestFactory(t, docsTestConfigWithAppID("docs-history-status"))
reg.Register(&httpmock.Stub{
Method: "GET",
URL: "/open-apis/docs_ai/v1/documents/doxcnHistory/history/revert_status",
Body: map[string]interface{}{
"code": 0,
"msg": "ok",
"data": map[string]interface{}{
"status": "partial_failed",
"history_version_id": "11",
"failed_block_tokens": []interface{}{"blk_1"},
},
},
})
err := mountAndRunDocs(t, DocsHistoryRevertStatus, []string{
"+history-revert-status",
"--doc", "doxcnHistory",
"--task-id", "task_1",
"--as", "bot",
}, f, stdout)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
data := decodeDocsHistoryEnvelope(t, stdout)
if got := data["status"]; got != "partial_failed" {
t.Fatalf("status = %#v, want partial_failed", got)
}
if got := data["history_version_id"]; got != "11" {
t.Fatalf("history_version_id = %#v, want 11", got)
}
failed, _ := data["failed_block_tokens"].([]interface{})
if len(failed) != 1 || failed[0] != "blk_1" {
t.Fatalf("failed_block_tokens = %#v, want [blk_1]", data["failed_block_tokens"])
}
}
func newDocsHistoryRuntimeCmd(t *testing.T, shortcut common.Shortcut, values map[string]string) *cobra.Command {
t.Helper()
cmd := &cobra.Command{Use: shortcut.Command}
for _, flag := range shortcut.Flags {
switch flag.Type {
case "int":
cmd.Flags().Int(flag.Name, 0, flag.Desc)
default:
cmd.Flags().String(flag.Name, flag.Default, flag.Desc)
}
}
for name, value := range values {
if err := cmd.Flags().Set(name, value); err != nil {
t.Fatalf("set --%s: %v", name, err)
}
}
return cmd
}
func decodeDocsHistoryEnvelope(t *testing.T, stdout *bytes.Buffer) map[string]interface{} {
t.Helper()
var envelope map[string]interface{}
if err := json.Unmarshal(stdout.Bytes(), &envelope); err != nil {
t.Fatalf("decode envelope: %v\nraw=%s", err, stdout.String())
}
data, _ := envelope["data"].(map[string]interface{})
if data == nil {
t.Fatalf("missing data in envelope: %#v", envelope)
}
return data
}
func TestDocsHistoryURLValidationMessage(t *testing.T) {
t.Parallel()
_, err := parseDocsHistoryDocRef("https://example.feishu.cn/doc/old_doc", "+history-list")
if err == nil {
t.Fatal("expected error")
}
if !strings.Contains(err.Error(), "only supports docx documents") {
t.Fatalf("unexpected error: %v", err)
}
}

View File

@@ -31,6 +31,8 @@ func docsSkillReadCommandForShortcut(shortcut string) string {
return docsSkillReadCommand + " references/lark-doc-fetch.md"
case "update":
return docsSkillReadCommand + " references/lark-doc-update.md"
case "history-list", "history-revert", "history-revert-status":
return docsSkillReadCommand + " references/lark-doc-history.md"
default:
return docsSkillReadCommand
}
@@ -44,6 +46,12 @@ func docsHelpCommandForShortcut(shortcut string) string {
return "lark-cli docs +fetch --help"
case "update":
return "lark-cli docs +update --help"
case "history-list":
return "lark-cli docs +history-list --help"
case "history-revert":
return "lark-cli docs +history-revert --help"
case "history-revert-status":
return "lark-cli docs +history-revert-status --help"
default:
return "lark-cli docs --help"
}
@@ -56,6 +64,9 @@ func Shortcuts() []common.Shortcut {
DocsCreate,
DocsFetch,
DocsUpdate,
DocsHistoryList,
DocsHistoryRevert,
DocsHistoryRevertStatus,
DocMediaInsert,
DocMediaUpload,
DocMediaPreview,

View File

@@ -8,6 +8,7 @@ import (
"encoding/json"
"fmt"
"path/filepath"
"strconv"
"strings"
"time"
@@ -28,6 +29,8 @@ const (
driveImport500MBFileSizeLimit int64 = 500 * 1024 * 1024
driveImport600MBFileSizeLimit int64 = 600 * 1024 * 1024
driveImport800MBFileSizeLimit int64 = 800 * 1024 * 1024
driveImportConcurrentOperationHint = "This import conflict means another operation is running in the same Drive location. Run batch imports to the same folder/root or target bitable serially. Wait a few seconds before retrying each failed import; retry each failed item at most 3 times, then stop and report the conflict."
)
// driveImportExtToDocTypes defines which source file extensions can be imported
@@ -47,6 +50,8 @@ var driveImportExtToDocTypes = map[string][]string{
"pptx": {"slides"},
}
var driveImportConcurrentOperationCodes = []int{232140101, 232140100, 233523001}
// driveImportSpec contains the user-facing import inputs after normalization.
type driveImportSpec struct {
FilePath string
@@ -427,11 +432,7 @@ func pollDriveImportTask(runtime *common.RuntimeContext, ticket string) (driveIm
return status, true, nil
}
if status.Failed() {
msg := strings.TrimSpace(status.JobErrorMsg)
if msg == "" {
msg = status.StatusLabel()
}
return status, false, errs.NewAPIError(errs.SubtypeServerError, "import failed with status %d: %s", status.JobStatus, msg)
return status, false, driveImportFailureError(status)
}
}
if !hadSuccessfulPoll && lastErr != nil {
@@ -440,3 +441,40 @@ func pollDriveImportTask(runtime *common.RuntimeContext, ticket string) (driveIm
return lastStatus, false, nil
}
func driveImportFailureError(status driveImportStatus) *errs.APIError {
msg := strings.TrimSpace(status.JobErrorMsg)
if msg == "" {
msg = status.StatusLabel()
}
apiErr := errs.NewAPIError(errs.SubtypeServerError, "import failed with status %d: %s", status.JobStatus, msg)
if code, ok := driveImportConcurrentOperationCode(msg); ok {
apiErr = apiErr.WithCode(code).WithRetryable().WithHint(driveImportConcurrentOperationHint)
}
return apiErr
}
func driveImportConcurrentOperationCode(msg string) (int, bool) {
for _, code := range driveImportConcurrentOperationCodes {
codeText := strconv.Itoa(code)
for idx := strings.Index(msg, codeText); idx >= 0; {
end := idx + len(codeText)
if (idx == 0 || !isASCIIDigit(msg[idx-1])) && (end == len(msg) || !isASCIIDigit(msg[end])) {
return code, true
}
nextStart := idx + 1
next := strings.Index(msg[nextStart:], codeText)
if next < 0 {
break
}
idx = nextStart + next
}
}
return 0, false
}
func isASCIIDigit(ch byte) bool {
return ch >= '0' && ch <= '9'
}

View File

@@ -7,6 +7,7 @@ import (
"bytes"
"errors"
"os"
"strconv"
"strings"
"testing"
@@ -211,6 +212,82 @@ func TestDriveImportStatusPendingWithoutToken(t *testing.T) {
}
}
func TestDriveImportFailureErrorAddsConcurrentOperationGuidance(t *testing.T) {
t.Parallel()
for _, code := range driveImportConcurrentOperationCodes {
t.Run(strconv.Itoa(code), func(t *testing.T) {
t.Parallel()
err := driveImportFailureError(driveImportStatus{
JobStatus: 3,
JobErrorMsg: "call CreateObjNode return error code, code: " + strconv.Itoa(code) + ", message:",
})
problem, ok := errs.ProblemOf(err)
if !ok {
t.Fatalf("expected typed error, got %T", err)
}
if problem.Category != errs.CategoryAPI {
t.Fatalf("category = %q, want %q", problem.Category, errs.CategoryAPI)
}
if problem.Subtype != errs.SubtypeServerError {
t.Fatalf("subtype = %q, want %q", problem.Subtype, errs.SubtypeServerError)
}
if problem.Code != code {
t.Fatalf("code = %d, want %d", problem.Code, code)
}
if !problem.Retryable {
t.Fatal("expected retryable error")
}
if problem.Hint != driveImportConcurrentOperationHint {
t.Fatalf("hint = %q, want %q", problem.Hint, driveImportConcurrentOperationHint)
}
})
}
}
func TestDriveImportFailureErrorLeavesOtherFailuresUnchanged(t *testing.T) {
t.Parallel()
tests := []struct {
name string
msg string
}{
{
name: "ordinary failure",
msg: "unsupported conversion",
},
{
name: "longer numeric code containing known code",
msg: "call CreateObjNode return error code, code: 12321401012, message:",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
t.Parallel()
err := driveImportFailureError(driveImportStatus{
JobStatus: 3,
JobErrorMsg: tt.msg,
})
problem, ok := errs.ProblemOf(err)
if !ok {
t.Fatalf("expected typed error, got %T", err)
}
if problem.Code != 0 {
t.Fatalf("code = %d, want 0", problem.Code)
}
if problem.Retryable {
t.Fatal("expected non-concurrency failure to remain non-retryable")
}
if problem.Hint != "" {
t.Fatalf("hint = %q, want empty", problem.Hint)
}
})
}
}
func TestDriveImportTimeoutReturnsFollowUpCommand(t *testing.T) {
f, stdout, _, reg := cmdutil.TestFactory(t, driveTestConfig())
reg.Register(&httpmock.Stub{

View File

@@ -170,6 +170,27 @@ func TestRegisterShortcutsMountsDocsMediaPreview(t *testing.T) {
}
}
func TestRegisterShortcutsMountsDocsHistoryCommands(t *testing.T) {
program := &cobra.Command{Use: "root"}
RegisterShortcuts(program, newRegisterTestFactory(t))
for _, name := range []string{"+history-list", "+history-revert", "+history-revert-status"} {
cmd, _, err := program.Find([]string{"docs", name})
if err != nil {
t.Fatalf("find docs %s shortcut: %v", name, err)
}
if cmd == nil || cmd.Name() != name {
t.Fatalf("docs %s shortcut not mounted: %#v", name, cmd)
}
if cmd.Flags().Lookup("api-version") != nil {
t.Fatalf("docs %s should not expose --api-version", name)
}
if !strings.Contains(cmd.Long, "lark-cli skills read lark-doc references/lark-doc-history.md") {
t.Fatalf("docs %s help missing history skill guidance:\n%s", name, cmd.Long)
}
}
}
func TestRegisterShortcutsDocsHelpAddsSkillReadGuidance(t *testing.T) {
program := &cobra.Command{Use: "root"}
RegisterShortcuts(program, newRegisterTestFactory(t))

View File

@@ -6,6 +6,7 @@
- 用户要把本地 `.xlsx` / `.csv` / `.base` 导入成 Base / 多维表格 / bitable第一步必须使用 `lark-cli drive +import --type bitable`
- 用户要把本地 `.md` / `.docx` / `.doc` / `.txt` / `.html` 导入成在线文档,使用 `lark-cli drive +import --type docx`
- 用户要把本地 `.xlsx` / `.xls` / `.csv` 导入成电子表格,使用 `lark-cli drive +import --type sheet`
- 批量执行 `drive +import` 且目标是同一个位置(同一 `--folder-token`、默认根目录,或同一 `--target-token`)时,必须串行执行;不要并发导入到同一位置,服务端可能返回并发冲突错误。
- 用户要在云空间里新建文件夹,优先使用 `lark-cli drive +create-folder`
- `lark-base` 只负责导入完成后的 Base 内部操作(表、字段、记录、视图),不要在“本地文件 -> Base”这一步提前切到 `lark-base`
@@ -194,6 +195,7 @@ lark-cli drive file.comments list --params '{"file_token": "xxx", "file_type": "
| `not exist` | 使用了错误的 token | 检查 token 类型wiki 链接必须先查询获取 `obj_token` |
| `permission denied` | 没有相关操作权限 | 引导用户检查当前身份对文档/文件是否有相应操作权限;如果需要,可以授予相应权限 |
| `invalid file_type` | file_type 参数错误 | 根据 `obj_type` 传入正确的 file_typedocx/doc/sheet/slides/bitable |
| `232140101` / `232140100` / `233523001`(常见于 `drive +import` 的 `job_error_msg` | 同一位置下存在并发导入 / 创建操作 | 批量导入到同一文件夹、根目录或同一 `--target-token` 时改为串行执行;每个失败项每次重试前等待几秒,总共最多重试 3 次,仍失败就停止并报告冲突 |
### 授权当前应用访问文档

View File

@@ -45,13 +45,15 @@ lark-cli calendar +agenda --as user
| 场景 | 前置要求 |
|------|----------|
| 预约日程/会议、查会议室 | 先读 [lark-calendar-schedule-meeting.md](references/lark-calendar-schedule-meeting.md) |
| 编辑已有日程 | 先定位目标日程 `event_id`;若是重复性日程,必须定位到具体实例的 `event_id`(禁止使用原重复日程 ID |
| 编辑已有日程 | 先定位目标日程 `event_id` |
| 编辑/删除重复性日程 | 先读 [重复性日程操作规范](references/lark-calendar-recurring.md),按操作范围(仅此次/全部/此次及后续)执行 |
| 删除/修改后验证 | 等待 2 秒再查询API 最终一致性),不要告知用户你等待了 |
| 调用任何 Shortcut | 先读其对应 reference 文档 |
## 核心概念
- **日程实例Instance**:重复性日程展开后的具体时间实例。操作重复日程的某次实例时,必须先定位该实例的 `event_id`,禁止使用原重复日程 `event_id`
- **日程实例Instance**:重复性日程展开后的具体时间实例。「仅此次」操作时使用具体实例的 `event_id`;「全部」或「此次及后续」操作时需对原重复性日程操作(使用原日程 `event_id`),并按需处理例外
- **重复性日程例外Exception**:对重复性日程某次实例做过「仅此次」编辑后产生的独立日程(拥有独立 `event_id`)。删除/更新「全部」时必须同时处理例外,否则例外会残留。
- **全天日程All-day Event**:只按日期占用、没有具体起止时刻的日程,结束日期是包含在日程时间内的。
- **时间块 vs 时间范围**:时间块是具体确定的连续时间段(如 `14:00~15:00`),时间范围是泛指(如"今天下午")。`+room-find` 必须基于确定时间块,不能基于模糊范围。
- **会议室Room**"room"不是"房间",是"会议室"。会议室是日程的一种参与人resource attendee不能脱离日程单独预定。
@@ -71,6 +73,7 @@ lark-cli calendar +agenda --as user
| 从日程获取关联的视频会议 ID 或用户绑定的会议纪要文档 | 本 skill`+meeting` |
| 从日程进一步拿 AI 智能纪要 / 逐字稿 / 妙记产物 | 先 `+meeting``meeting_id`,再 [`vc +detail`](../lark-vc/references/lark-vc-detail.md) → [`note +detail`](../lark-note/references/lark-note-detail.md) / [`minutes +detail`](../lark-minutes/references/lark-minutes-detail.md) |
| 预约/改约日程、添加/移除参会人、添加/更换会议室、调整时间 | 先判断新建 vs 编辑,再进入 [schedule-meeting 工作流](references/lark-calendar-schedule-meeting.md) |
| 编辑/删除重复性日程(「改这个重复日程」「删掉后面的」「全部取消」等) | 先读 [重复性日程操作规范](references/lark-calendar-recurring.md),确认操作范围后执行 |
## 任务类型分流

View File

@@ -47,6 +47,7 @@ lark-cli calendar +create --summary "..." --start "..." --end "..." \
> 自动设置 `reminders: [{"minutes": 5}]`,默认日程开始前 5 分钟提醒。
> 自动设置 `vchat: {"vc_type": "vc"}`,默认日程包含飞书视频会议。如需其他视频会议类型或不含视频会议,请使用完整 API 命令。
> 失败保护:若添加参会人失败(如 open_id 错误CLI 会自动删除刚创建的空日程(回滚,不通知参会人)。
> 审批会议室:`+create` 不暴露低频字段 `attendees[].approval_reason`。如果会议室要求审批,请使用用户身份先创建日程,再用完整 API `calendar event.attendees create --as user` 添加会议室并传 `approval_reason`。
## 高级用法(完整 API 命令)
@@ -72,9 +73,16 @@ lark-cli calendar events create \
lark-cli schema calendar.event.attendees.create
## 添加参会人
lark-cli calendar event.attendees create \
--as user \
--params '{"calendar_id":"<CALENDAR_ID>","event_id":"<EVENT_ID>"}' \
--data '{"attendees": [{"type": "user", "user_id": "ou_xxx"}]}'
## 添加需要审批的会议室approval_reason 最大 200 字符)
lark-cli calendar event.attendees create \
--as user \
--params '{"calendar_id":"<CALENDAR_ID>","event_id":"<EVENT_ID>"}' \
--data '{"attendees": [{"type": "resource", "room_id": "omm_xxx", "approval_reason": "申请原因"}]}'
# 可选第三步(推荐):若第二步失败,回滚删除空日程
## 查看完整参数定义
lark-cli schema calendar.events.delete

View File

@@ -0,0 +1,90 @@
# 重复性日程操作规范
重复性日程的编辑/删除分为三种范围:「仅此次」「全部」「此次及后续」。用户未明确范围时,**必须询问确认**。
## 关键概念
- **event_id 结构**`event_id` 的格式为 `{event_uid}_{originalTime}`。普通日程或重复性日程本体的 `originalTime``0`;例外的 `originalTime > 0`,代表该例外在原重复性序列中本来的时间位置。因此 `{event_uid}_0` 即为原重复性日程的 `event_id`
- **原重复性日程**:携带 `rrule` 的日程本体,`event_id` 形如 `{event_uid}_0`。系列的所有属性标题、时间、rrule、描述等都挂在本体上。
- **例外Exception**:对某次实例做过「仅此次」编辑后产生的独立日程,`event_id` 形如 `{event_uid}_{originalTime}``originalTime > 0`)。通过 `event_uid` 部分即可关联回原重复性日程。
- 删除/更新原重复性日程 **不会** 级联处理例外——必须手动逐个处理。
## 前置步骤(所有范围通用)
1. 通过 `+agenda``+search-event` 定位重复性日程,获取原重复性日程的 `event_id`
2. 通过 `events instance_view``+agenda` 列出实例,识别哪些是例外(`event_id``originalTime > 0` 的即为例外)。
3. 确认用户的操作范围。
## 编辑全部(更新时间)
| 步骤 | 命令 | 说明 |
|------|------|------|
| 1 | `lark-cli calendar +update --event-id <原重复日程ID> --start ... --end ...` | 更新原重复性日程的时间 |
| 2 | `lark-cli calendar events delete --params '{"calendar_id":"<CAL_ID>","event_id":"<例外ID>","need_notification":false}'` (逐个) | 时间变更后例外已无意义,必须删除 |
> 理由:更新时间会改变重复起止点,例外日程的原始占位已变,若保留会导致时间冲突或残留。
## 编辑全部(更新非时间字段)
| 步骤 | 命令 | 说明 |
|------|------|------|
| 1 | `lark-cli calendar +update --event-id <原重复日程ID> --summary ... --description ...` | 更新原重复性日程的标题/描述等 |
| 2 | `lark-cli calendar +update --event-id <例外ID> --summary ... --description ...` (逐个) | 同步更新例外日程的对应字段 |
> 理由:例外已脱离原重复性日程独立存在,不会自动继承原日程的更新。
## 删除全部
| 步骤 | 命令 | 说明 |
|------|------|------|
| 1 | `lark-cli calendar events delete --params '{"calendar_id":"<CAL_ID>","event_id":"<原重复日程ID>","need_notification":true}'` | 删除重复性日程本体 |
| 2 | `lark-cli calendar events delete --params '{"calendar_id":"<CAL_ID>","event_id":"<例外ID>","need_notification":false}'` (逐个) | 删除所有例外日程 |
> 理由:例外是独立实体,删除原重复性日程不会级联删除例外。
## 编辑此次及后续
| 步骤 | 命令 | 说明 |
|------|------|------|
| 1 | `lark-cli calendar +update --event-id <原重复日程ID> --rrule "FREQ=...;UNTIL=<截止日期>"` | 截短原重复性日程UNTIL 设为指定时间前一次实例的日期) |
| 2 | `lark-cli calendar events delete ...` (逐个) | 删除指定时间之后(含)的例外日程 |
| 3 | `lark-cli calendar +create --summary ... --start <指定时间> --end ... --rrule "FREQ=..." --attendee-ids ...` | 从指定时间开始创建新的重复性日程(即「后续」部分,携带编辑后的内容) |
> UNTIL 计算规则:若用户选择「从第 N 次开始编辑」UNTIL 应设置为第 N-1 次实例的日期(即保留到指定时间之前的最后一次)。
> 新日程应继承原日程的参会人、会议室等配置(除非用户明确要修改)。
## 删除此次及后续
| 步骤 | 命令 | 说明 |
|------|------|------|
| 1 | `lark-cli calendar +update --event-id <原重复日程ID> --rrule "FREQ=...;UNTIL=<截止日期>"` | 截短原重复性日程UNTIL 设为指定时间前一次实例的日期) |
| 2 | `lark-cli calendar events delete ...` (逐个) | 删除指定时间之后(含)的例外日程 |
> 与「编辑此次及后续」的区别:不需要步骤 3创建新的重复性日程因为目标是删除后续而非替换。
## 仅此次
- **编辑仅此次**:通过 `+agenda` / `+search-event` 定位到具体实例的 `event_id`,然后正常调用 `+update`
- **删除仅此次**:定位到具体实例的 `event_id`,调用 `events delete`
## 用户意图映射
| 用户表达 | 操作范围 |
|----------|----------|
| 「改这个重复日程的标题」「全部改」「每次都改」 | 编辑全部 |
| 「删掉这个重复日程」「取消所有」 | 删除全部 |
| 「从下周开始改时间」「后面的都改」 | 编辑此次及后续 |
| 「从下周开始不要了」「后面的都删」 | 删除此次及后续 |
| 「就改这一次」「只删这一次」 | 仅此次 |
| 未明确范围 | **必须询问用户** |
## 注意事项
- 涉及时间戳计算(如推算 UNTIL 日期)时,必须调用系统命令或脚本,禁止心算。
## 参考
- [lark-calendar](../SKILL.md) — 日历全部命令
- [lark-calendar-update](lark-calendar-update.md) — 更新日程 Shortcut
- [lark-calendar-create](lark-calendar-create.md) — 创建日程 Shortcut
- [lark-shared](../../lark-shared/SKILL.md) — 认证和全局参数

View File

@@ -43,7 +43,7 @@ lark-cli calendar +update \
| 参数 | 必填 | 说明 |
|------|------|------|
| `--event-id <id>` | 是 | 要更新的日程 ID。重复性日程要先定位到目标实例的 `event_id`,不要直接使用原重复日程 ID |
| `--event-id <id>` | 是 | 要更新的日程 ID。重复性日程请根据操作范围选择 ID详见 [重复性日程操作规范](lark-calendar-recurring.md) |
| `--calendar-id <id>` | 否 | 日历 ID省略则使用 `primary` |
| `--summary <text>` | 否 | 新日程标题。仅在显式传入 `--summary` 时更新;若传空字符串,会把标题清空 |
| `--description <text>` | 否 | 新日程描述。目前 API 方式不支持编辑富文本描述;如果日程描述通过客户端编辑为富文本内容,则使用 API 更新描述会导致富文本格式丢失。仅在显式传入 `--description` 时更新;若传空字符串,会把描述清空 |
@@ -65,7 +65,7 @@ lark-cli calendar +update \
- 只想修改标题、描述、时间或重复规则时,不需要同时传 `--add-attendee-ids``--remove-attendee-ids`
- 如需替换某个参与人、群组或会议室,使用 `--remove-attendee-ids <旧ID>` + `--add-attendee-ids <新ID>`
- 会议室是 resource attendee必须使用 `omm_` ID 添加到参会人列表,不能脱离日程单独预定。
- 更新重复性日程的某一次实例时,必须先通过 `+agenda``+search-event` 或实例视图定位该实例的 `event_id`
- 更新重复性日程时,必须先确定操作范围(仅此次/全部/此次及后续),然后按 [重复性日程操作规范](lark-calendar-recurring.md) 执行
- 如果需要验证更新结果,等待至少 2 秒后再查询,避免同步延迟导致读到旧数据。
- 当同一次命令组合多个动作时,执行顺序为“日程字段 -> 移除参会人 -> 添加参会人”。若中途失败,不会自动回滚已成功步骤;错误信息会说明已完成的步骤。

View File

@@ -5,7 +5,7 @@ description: "飞书云文档Docx / Wiki 文档):读取和编辑飞书文
metadata:
requires:
bins: ["lark-cli"]
cliHelp: "lark-cli docs --help; lark-cli docs +create --help; lark-cli docs +fetch --help; lark-cli docs +update --help; lark-cli docs +resource-download --help; lark-cli docs +resource-update --help; lark-cli docs +resource-delete --help; lark-cli mindnotes nodes list --help; lark-cli mindnotes nodes create --help"
cliHelp: "lark-cli docs --help;lark-cli mindnotes --help"
---
# docs
@@ -21,41 +21,28 @@ lark-cli docs +update --doc "文档URL或token" --command append --content '<p>
## 前置条件 — 执行操作前必读
**CRITICAL — 按任务命中以下阅读规则;执行对应操作前MUST 先用 Read 工具读取命中的文件,缺一不可:**
**CRITICAL — 执行对应操作前MUST 先用 Read 工具读取以下文件,缺一不可:**
1. [`../lark-shared/SKILL.md`](../lark-shared/SKILL.md) — 认证、权限处理、全局参数(所有操作通用)
2. **读取文档(`docs +fetch`** → 必读 [`lark-doc-fetch.md`](references/lark-doc-fetch.md)`--scope` / `--detail` 选择、局部读取策略、`<fragment>` / `<excerpt>` 输出结构)
3. **创建或编辑文档内容** → 必读 [`lark-doc-xml.md`](references/lark-doc-xml.md)XML 语法规则,仅当用户明确要求 Markdown 时改读 [`lark-doc-md.md`](references/lark-doc-md.md))和必读 [`lark-doc-style.md`](references/style/lark-doc-style.md)(写作原则:默认段落、按体裁、组件克制);从零创建时加读 [`lark-doc-create-workflow.md`](references/style/lark-doc-create-workflow.md);编辑已有文档时加读 [`lark-doc-update.md`](references/lark-doc-update.md) 和 [`lark-doc-update-workflow.md`](references/style/lark-doc-update-workflow.md)
1. **所有文档操作** → 必读 [`../lark-shared/SKILL.md`](../lark-shared/SKILL.md)
用于认证、权限处理、身份选择和全局参数。
2. **读取文档(`docs +fetch`** → 必读 [`lark-doc-fetch.md`](references/lark-doc-fetch.md)
用于选择 `--scope` / `--detail`、局部读取策略,以及理解 `<fragment>` / `<excerpt>` 输出结构。
3. **创建、重写、润色、排版或生成较多正文** → 必读:
- [`lark-doc-design-philosophy.md`](references/lark-doc-design-philosophy.md) — 文档设计判断:读者任务、结构取舍、组件选择、体裁路由和反模板化。
- [`lark-doc-xml.md`](references/lark-doc-xml.md) — XML 语法规则;仅当用户明确要求 Markdown、提供 `.md` 文件,或明确要求导入 Markdown 时,改读 [`lark-doc-md.md`](references/lark-doc-md.md)。
进一步按场景加读:
- 如果 [`lark-doc-design-philosophy.md`](references/lark-doc-design-philosophy.md) 的体裁路由命中某个 `references/genres/*.md` 文件,只读取该体裁文件;不要读取整个 `genres/` 目录。
- 从零创建文档时,加读 [`lark-doc-create.md`](references/lark-doc-create.md)。
- 编辑已有文档时,加读 [`lark-doc-update.md`](references/lark-doc-update.md)。
4. **仅做微编辑**,例如明确旧文本 → 新文本的简单替换、错别字修正、日期 / 人名 / 数字替换,不需要读取设计哲学和体裁文件;按 [`lark-doc-update.md`](references/lark-doc-update.md) 执行,并在写后 fetch 验证。
**Philosophy 负责判断Genre 负责偏好Create / Update 负责执行XML / Markdown 负责合法表达。**
**未读完命中的必读文件就执行相应操作,容易导致参数选择错误、格式错误、结构失焦或内容破坏。**
**未读完以上文件就执行相应操作会导致参数选择错误或格式错误。**
> **格式选择规则(全局):**
> - **创建 / 导入场景**`docs +create`,或 `docs +update --command append/overwrite` 的整段写入XML 和 Markdown 都可以。用户提供 `.md` 本地文件、或明确说"导入 Markdown"时,直接用 Markdown否则默认 XML(可用 callout、grid、checkbox 等富 block
> - **创建 / 导入场景**`docs +create`,或 `docs +update --command append/overwrite` 的整段写入XML 和 Markdown 都可以。用户提供 `.md` 本地文件、或明确说"导入 Markdown"时,直接用 Markdown否则默认 XML。
> - **精准编辑场景**`docs +update` 的 `str_replace` / `block_insert_after` / `block_replace` / `block_delete` / `block_move_after` 等局部精修指令):优先使用 XML`--doc-format xml`即默认值。XML 能稳定表达 block 结构和样式,局部精修更可控;不要因为 Markdown 更简单就自行切换。
## 快速决策
- 用户要**复制文档 / 创建文档副本 / 另存为副本**时,切到 [`lark-drive`](../lark-drive/SKILL.md),按其中的复制指引使用 `lark-cli drive files copy`;不要用 `docs +fetch` + `docs +create` 重建正文,也不要走 `drive +export` / `drive +import`
- 先判定任务路径:找文档 / 导入导出走 [`lark-drive`](../lark-drive/SKILL.md);只读 / 摘要用 `docs +fetch` 默认 `simple`已有文档改写按 [`lark-doc-update.md`](references/lark-doc-update.md) 的 Observe-Diagnose-Patch Loop 先 fetch 再局部 patch明确旧文本 → 新文本的简单替换可直接 `str_replace`,但写后必须 fetch 验证;只有 block 链接、评论锚点、插入 / 替换 / 删除 / 移动才局部 fetch `with-ids`;保真改写已有内容才读 `full`
- 先判定任务路径:找文档 / 导入导出走 [`lark-drive`](../lark-drive/SKILL.md);只读 / 摘要用 `docs +fetch` 默认 `simple`;明确旧文本 → 新文本直接 `str_replace`;只有 block 链接、评论锚点、插入 / 替换 / 删除 / 移动才局部 fetch `with-ids`;保真改写已有内容才读 `full`
- block 直达链接格式:`文档基础 URL#block_id`;没有 block_id 时局部 fetch `with-ids`
- 连续执行多个文档写操作时,必须按 [`lark-doc-update.md`](references/lark-doc-update.md) 的「Block ID 生命周期」处理:每次更新后都按 block ID 已变更处理;需要继续或重复修改时,先重新 fetch 最新内容和 block ID不要复用旧 fetch 结果
- 连续执行多个文档写操作时,必须按 [`lark-doc-update.md`](references/lark-doc-update.md) 的「Block ID 生命周期」判断旧 block ID 是否还能复用;`overwrite` / `block_replace` / `block_delete` 后不要复用受影响的旧 ID插入 / 复制后要重新 fetch 才能拿到新 block ID
- 用户需要在文档内**创建、复制或移动**资源块(画板、电子表格、多维表格等)时,必须先读取 [`lark-doc-xml.md`](references/lark-doc-xml.md) 的「三、资源块」章节
- 写文档时,由内容和用户意图决定表达形式;流程、架构、路线图、关键指标等信息可以使用画板,但不要默认把重要信息都画板化
- 新增画板按复杂度处理:简单 Mermaid / SVG 图可由主 Agent 直接写入草稿;复杂图或需要专门视觉设计的 SVG 交给 SubAgent 产出完整 `<whiteboard type="svg">...</whiteboard>`;特别复杂或已有画板更新,主 Agent 先建 `<whiteboard type="blank"></whiteboard>`,再启动 SubAgent 读取 `lark-whiteboard` 写入
- 新增或更新画板时,按 [`lark-doc-whiteboard.md`](references/lark-doc-whiteboard.md) 选型Mermaid 可由主 Agent 直接插入SVG / 复杂图 / 已有画板更新按其中流程隔离到 SubAgent
- 用户说"看一下文档里的图片/附件/素材""预览素材" → 用 `lark-cli docs +media-preview`
- 用户明确说"下载素材" → 用 `lark-cli docs +media-download`
- 用户想把文档回滚到某个 `revision_id` 或某一时刻 → 先读 [`lark-doc-history.md`](references/lark-doc-history.md),按其中流程操作
- 用户明确说"下载/更新/删除文档封面图" → 用 `lark-cli docs +resource-download/+resource-update/+resource-delete --type cover`
- `resource-*` 目前仅支持 Docx 封面资源;其他图片、附件或素材请走 `+media-*`
- 如果目标是画板/whiteboard/画板缩略图 → 只能用 `lark-cli docs +media-download --type whiteboard`(不要用 `+media-preview`
@@ -83,6 +70,7 @@ Shortcut 是对常用操作的高级封装(`lark-cli docs +<verb> [flags]`
| [`+create`](references/lark-doc-create.md) | Create a Lark document (XML / Markdown) |
| [`+fetch`](references/lark-doc-fetch.md) | Fetch Lark document content (XML / Markdown / im-markdown; `im-markdown` only after fetch for `lark-im`) |
| [`+update`](references/lark-doc-update.md) | Update a Lark document (str_replace / block_insert_after / block_replace / ...) |
| [`+history-list` / `+history-revert` / `+history-revert-status`](references/lark-doc-history.md) | List document history, revert to a `history_version_id`, and query revert task status |
| [`+media-insert`](references/lark-doc-media-insert.md) | Insert a local image or file at the end of a Lark document (4-step orchestration + auto-rollback). Prefer `--from-clipboard` when the image is already on the system clipboard (screenshots, copy from Feishu/browser); use `--file` only for on-disk sources. |
| [`+media-download`](references/lark-doc-media-download.md) | Download document media or whiteboard thumbnail (auto-detects extension) |
| [`+media-preview`](references/lark-doc-media-preview.md) | Preview document media file (auto-detects extension) |

View File

@@ -1,38 +0,0 @@
# Genre: Formal Document
适用:通知、制度、公告、汇报、立项、正式方案、对外说明、需要归档或跨团队协作的文档。
## Structure patterns
按任务选择结构,不要机械套用:
- **事项通知**:事项 → 适用范围 → 具体安排 → 时间节点 → 责任人 / 联系方式。
- **决策汇报**:结论 / 建议 → 背景 → 方案对比 → 风险影响 → 待决策问题。
- **制度规范**:目的 → 适用范围 → 规则正文 → 例外情况 → 执行与解释。
- **正式方案**:问题定义 → 目标与范围 → 方案 → 资源 / 排期 → 风险 → 下一步。
## Style
准确、克制、明确责任边界。优先写清对象、动作、时间、范围和责任人。少用夸张修饰,不使用网感标题。避免用“赋能、闭环、抓手、沉淀、拉通、对齐”等空泛词替代具体动作。
## Component bias
- 时间、责任、范围、事项清单:优先 table。
- 风险、待确认、强约束:可用 callout。
- 方案对比table / grid / 决策矩阵。
- 流程、组织关系、里程碑:必要时用 whiteboard。
- 长期归档、对外说明:少用颜色和 emoji保持朴素。
## Avoid
- 开头铺太长背景,迟迟不给结论。
- 缺少责任人、时间、范围、生效条件。
- 为了正式感堆砌抽象词。
- 为了美观强行加入 callout、grid 或画板。
## Final check
- 事项、范围、责任、时间是否明确?
- 结论和待决策问题是否前置?
- 语气是否正式、克制、无夸张?
- 是否适合归档和后续追责?

View File

@@ -1,39 +0,0 @@
# Genre: PRD / Product Requirement
适用PRD、需求说明、用户故事、产品方案、功能设计、版本范围说明。
## Structure patterns
按任务选择结构:
- **标准 PRD**:背景与问题 → 用户与场景 → 目标与非目标 → 方案概述 → 功能范围 → 交互 / 流程 → 数据与埋点 → 边界异常 → 验收标准 → 风险依赖。
- **轻量需求**:问题 → 目标 → 用户故事 → 规则 → 验收标准 → 待确认。
- **产品方案**:现状 → 机会 / 问题 → 方案 → 取舍 → 里程碑 → 风险。
- **功能评审**:结论 → 范围变化 → 核心流程 → 争议点 → 待决策问题。
## Style
具体、可评审、可验收。多写用户行为、系统响应、状态变化和规则,少写抽象愿景。每个需求尽量落到状态、规则、字段、交互或验收条件。
## Component bias
- 用户故事、验收标准:列表 / checkbox。
- 字段、规则、状态、埋点table。
- 用户流程、系统流程、状态机whiteboard。
- 风险、依赖、待确认callout。
- 版本范围、需求优先级table / 小标题。
## Avoid
- 只有需求描述,没有验收标准。
- 混淆目标、方案、交互和实现细节。
- 没有非目标,导致范围无限扩大。
- 缺少异常状态、边界条件和数据规则。
- 用“大幅提升体验”替代具体用户收益。
## Final check
- 目标、非目标和范围是否清楚?
- 用户、场景、规则和状态是否可评审?
- 验收标准是否可验证?
- 风险、依赖和待确认项是否明确?

View File

@@ -1,39 +0,0 @@
# Genre: Retrospective / Summary
适用:项目复盘、事故复盘、阶段总结、周报、月报、季度总结、问题回顾。
## Structure patterns
按任务选择结构:
- **事故复盘**:结论 → 影响范围 → 时间线 → 根因 → 处置过程 → 暴露问题 → 改进动作 → 负责人 / 截止时间。
- **项目复盘**:目标 → 结果 → 关键事实 → 做得好的地方 → 问题与原因 → 经验沉淀 → 后续动作。
- **周报 / 月报**:核心进展 → 关键结果 → 风险阻塞 → 下阶段计划 → 需要支持。
- **阶段总结**:目标回顾 → 数据 / 事实 → 判断 → 问题 → 后续重点。
## Style
事实和判断分开。避免甩锅,也避免空泛反思。后续动作必须具体到负责人、时间和验收方式。不要用“加强对齐、持续优化、提升意识”替代真实动作。
## Component bias
- 时间线table / whiteboard。
- 根因和影响链路:列表 / 鱼骨图 / 因果图。
- 行动项checkbox / table。
- 风险和阻塞callout。
- 数据对比table / whiteboard 图表。
## Avoid
- 只有情绪评价,没有事实。
- 只有“后续加强”,没有负责人和时间。
- 时间线缺失,无法复盘过程。
- 结论与证据不匹配。
- 把复盘写成汇功或甩锅材料。
## Final check
- 事实、判断、动作是否分开?
- 时间线和影响范围是否清楚?
- 根因是否有证据支持?
- 后续动作是否有负责人、截止时间和验收标准?

View File

@@ -1,41 +0,0 @@
# Genre: Xiaohongshu / Social Content
适用:小红书笔记、爆文、种草、攻略、经验分享、标题优化、社媒传播型内容。
## Structure patterns
按任务选择结构:
- **经验型**:强钩子 → 痛点场景 → 个人经验 / 观察 → 方法清单 → 适用人群 → 互动结尾。
- **攻略型**:结果承诺 → 准备条件 → 步骤 → 避坑点 → 清单总结。
- **种草型**:使用场景 → 核心卖点 → 真实体验 → 对比 / 注意事项 → 行动建议。
- **观点型**:反常识判断 → 论据 → 例子 → 常见误区 → 行动建议。
- **清单型**:适用人群 → 清单分组 → 每项解释 → 收藏提醒 / 使用方式。
## Style
口语化、具体、有节奏。短段落优先,多用场景、例子和清单。可以有情绪,但不能编造经历、数据或效果。标题可以吸引人,但正文必须兑现标题承诺。
## Component bias
- 开头钩子:短段落 / callout。
- 方法步骤:列表 / checkbox。
- 避坑、对比、清单grid / table。
- 可收藏内容checkbox / table。
- 复杂流程才用 whiteboard轻内容不要图表化。
- emoji 可用,但要克制并保持语义一致。
## Avoid
- 标题党但正文兑现不了。
- 过度使用“绝了、封神、天花板、救命、狠狠爱住”等空泛网感词。
- 编造个人经历、效果数据、医学 / 金融 / 法律结论。
- 信息太散,没有可保存的方法、清单或判断。
- 为了网感牺牲事实准确性。
## Final check
- 标题是否有吸引力且不虚假?
- 开头 3 秒内是否说明痛点或利益?
- 正文是否有具体方法、例子或清单?
- 是否适合收藏、转发或执行?

View File

@@ -1,40 +0,0 @@
# Genre: SOP / Tutorial
适用SOP、操作手册、培训材料、入门指南、使用说明、FAQ、流程规范。
## Structure patterns
按任务选择结构:
- **SOP**:适用范围 → 前置条件 → 操作步骤 → 检查标准 → 异常处理 → 责任边界。
- **入门教程**:适合谁 → 准备什么 → 快速开始 → 分步说明 → 常见问题 → 下一步。
- **操作手册**:任务列表 → 每个任务的步骤 → 注意事项 → 错误处理 → 参考链接。
- **FAQ**:问题分类 → 问题 → 简短答案 → 详细说明 / 操作入口。
## Style
直接、清楚、面向操作者。一句话只承担一个任务。按用户看到的界面命名,不按系统内部实现命名。优先使用动词开头的步骤,例如“打开”“选择”“填写”“确认”。
## Component bias
- 步骤:有序列表。
- 检查项checkbox。
- 命令、配置、日志pre / code。
- 错误处理、条件分支table / callout。
- 流程复杂时whiteboard。
- 入口链接button / link / url-preview。
## Avoid
- 缺少前置条件。
- 步骤跳跃,默认读者知道上下文。
- 用内部术语命名用户界面。
- 没有失败处理或校验方式。
- 一个步骤里塞多个动作。
## Final check
- 新手能否按步骤完成任务?
- 前置条件、输入、输出和校验是否清楚?
- 异常情况是否有处理方式?
- 每一步是否只做一件事?

View File

@@ -1,40 +0,0 @@
# Genre: Technical Document
适用技术方案、架构设计、接口说明、API 文档、部署手册、迁移方案、排障指南、研发交接。
## Structure patterns
按任务选择结构:
- **技术方案**:问题 / 背景 → 目标与非目标 → 约束 → 总体设计 → 模块设计 → 数据 / 接口 → 风险 → 验证 → 灰度 / 回滚。
- **接口文档**:适用范围 → 鉴权 → 请求参数 → 响应结构 → 错误码 → 示例 → 兼容性。
- **排障指南**:现象 → 影响范围 → 快速判断 → 排查步骤 → 常见原因 → 修复方式 → 升级路径。
- **迁移方案**:范围 → 前置条件 → 步骤 → 校验 → 回滚 → 风险与窗口期。
- **架构文档**:目标 → 上下文 → 架构图 → 核心链路 → 模块职责 → 数据流 → 边界限制。
## Style
精确、朴素、可实现、可验证。术语保持一致,不为了通俗牺牲准确性。不确定处标注“假设”或“待确认”。不要把技术取舍写成宣传话术。
## Component bias
- 命令、代码、配置、日志pre / code。
- 参数、字段、错误码、兼容性table。
- 架构、调用链、状态流转、部署拓扑whiteboard / Mermaid。
- 操作步骤、验收项:列表 / checkbox。
- 风险、限制、回滚条件callout。
## Avoid
- 只有概念,没有接口、数据、边界或验证。
- 忽略失败场景、兼容性、监控、回滚。
- 图很多,但没有说明依赖、调用关系或数据流。
- 使用营销腔描述技术收益。
- 改写时破坏代码、图片、引用、资源块或 token 化内容。
## Final check
- 读者能否据此实现、集成、排查或评审?
- 目标、非目标、边界和失败场景是否清楚?
- 参数、命令、接口和示例是否准确?
- 是否有验证、灰度、回滚或排障路径?

View File

@@ -1,30 +1,15 @@
# docs +create创建飞书云文档
> **前置条件MUST READ** 生成文档内容前,必须先用 Read 工具读取以下文件,缺一不可:
> 1. [`lark-doc-design-philosophy.md`](lark-doc-design-philosophy.md) — 文档设计判断(读者任务、结构取舍、组件选择、反模板化
> 2. [`lark-doc-xml.md`](lark-doc-xml.md) — XML 语法规则(使用 Markdown 格式时改读 [`lark-doc-md.md`](lark-doc-md.md)
> 1. [`lark-doc-xml.md`](lark-doc-xml.md) — XML 语法规则(使用 Markdown 格式时改读 [`lark-doc-md.md`](lark-doc-md.md)
> 2. [`lark-doc-style.md`](style/lark-doc-style.md) — 写作原则(默认段落、按体裁、组件克制
> 3. [`lark-doc-create-workflow.md`](style/lark-doc-create-workflow.md) — 从零创作工作流Code-Act Loop、单 Agent 串行撰写)
>
> **未读完以上文件就生成内容会导致结构失焦或格式错误。**
> **未读完以上文件就生成内容会导致格式错误。**
从 XML默认或 Markdown 内容创建一个新的飞书云文档。
> **⚠️ 格式选择规则:** 创建 / 导入场景下 XML 和 Markdown 都可以——用户提供 `.md` 本地文件、或明确说"导入 Markdown"时,直接用 Markdown没有明确指示时默认 XML表达能力更强支持 callout、grid、checkbox 等富 block 类型)。不要在用户没要求的情况下主动从 XML 切到 Markdown也不要在用户已给出 Markdown 时强行改成 XML。
## Document Authoring Loop
从零创作文档时,以本地草稿文件作为唯一正文源(默认 XML用户明确要求 Markdown 或提供 `.md` 文件时使用 Markdown主 Agent 串行完成写作、审阅和修订,草稿达标后再调用 `docs +create`。不要先创建线上骨架再让多个 Agent 并行拆章节写入;这容易造成风格漂移、重复铺陈、结构断流和事实口径不一致。
1. **Plan规划**:明确受众、目的、范围、体裁、结构和验收标准;事实来源或产品口径不确定时先询问用户,能安全假设时说明假设并继续。
2. **Draft起草**:在当前工作目录创建相对路径草稿文件,主 Agent 按文档顺序串行写作;
3. **Observe观察**:读取当前草稿,检查结构、重复、断流、事实、语气和格式;
4. **Revise修订**:局部修订本地文件,优先调整顺序、补桥接句、删重复、统一术语和修正 XML不追求第一稿一次完美。
5. **Enrich增强**决定一篇文档是否真正好读、可信、可传播。必要时加入表格、列表、callout、grid、画板、引用或附件增强必须降低理解成本不为了“丰富”而装饰。
- 新增画板按复杂度处理:简单 Mermaid 或 SVG 图可由主 Agent 直接写入本地 XML 草稿;复杂 SVG 可启动 SubAgent 只产出完整 `<whiteboard type="svg">...</whiteboard>` 片段;特别复杂或已有画板更新,再创建空白画板并让 SubAgent 读取 `lark-whiteboard` 完成写入。
- 创建后才能插入的图片、附件或资源先在草稿中标记位置和素材路径Deliver 后用 `docs +media-insert` 等命令补齐并 fetch 验证。
6. **Validate验证**:检查用户要求、字数、格式、事实、可读性和风格;任何一项不满足,就回到 Observe / Revise 继续 loop。
7. **Deliver交付**:本地草稿通过验证后,用 `docs +create --content @lark-doc-draft.xml` 创建 XML 文档Markdown 草稿用 `docs +create --doc-format markdown --content @lark-doc-draft.md`。如需创建后插入图片、附件或资源,完成后再次 fetch 确认。
`@file` 只接受当前工作目录下的相对路径,不要传 `@/tmp/xxx.xml` 这类绝对路径。SubAgent 只用于边界清晰的视觉或资源片段(如完整 SVG whiteboard XML主 Agent 必须审核并合并结果;不要让 SubAgent 并行撰写正文。
> **⚠️ 格式选择规则:** 创建 / 导入场景下 XML 和 Markdown 都可以——用户提供 `.md` 本地文件、或明确说"导入 Markdown"时,直接用 Markdown没有明确指示时默认 XML表达能力更强可承载更丰富的结构化内容)。不要在用户没要求的情况下主动从 XML 切到 Markdown也不要在用户已给出 Markdown 时强行改成 XML。
## 命令
@@ -80,9 +65,15 @@ lark-cli docs +create --doc-format markdown --title "项目计划" --content $'#
| `--parent-token` | 否 | 父文件夹或知识库节点 token`--parent-position` 互斥) |
| `--parent-position` | 否 | 父节点位置,如 `my_library`(与 `--parent-token` 互斥) |
## 最佳实践
- **较长文档**:参考 [`lark-doc-create-workflow.md`](style/lark-doc-create-workflow.md) 先建骨架再分段写入;短文档可一次写完整内容
- **表达形式**:由用户目标和内容决定。需要结构化表达时可参考 [`lark-doc-style.md`](style/lark-doc-style.md),但不要默认套用固定开头、固定富 block 比例或固定图表
## 参考
- [`lark-doc-design-philosophy.md`](lark-doc-design-philosophy.md) — 文档设计判断与组件取舍
- [`lark-doc-create-workflow.md`](style/lark-doc-create-workflow.md) — 从零创作工作流Code-Act Loop、单 Agent 串行撰写)
- [`lark-doc-style.md`](style/lark-doc-style.md) — 文档写作原则(默认段落、按体裁、组件克制)
- [`lark-doc-xml.md`](lark-doc-xml.md) — XML 语法规范
- [`lark-doc-fetch.md`](lark-doc-fetch.md) — 获取文档
- [`lark-doc-update.md`](lark-doc-update.md) — 更新文档

View File

@@ -1,98 +0,0 @@
# Lark Doc Design Philosophy
- 把飞书云文档当作帮助读者完成理解、判断、决策或行动的界面,而不是内容容器。你的任务不是把信息铺满页面,而是为具体读者设计一条清晰、可信、可执行的阅读路径。
- 本文件负责文档设计判断:读者任务、结构取舍、表达组件、反模板化和质量自检。具体执行流程见 `lark-doc-create.md``lark-doc-update.md`XML / Markdown 语法见 `lark-doc-xml.md` / `lark-doc-md.md`;体裁结构和语气差异按下方路由读取 `genres/*.md`
## 0. Priority order
当目标冲突时,按此顺序取舍:**事实准确性 > 信息层级 > 可扫描性 > 协作可执行性 > 视觉整洁 > 个性化表达**。事实不确定就标注假设 / 待确认;结构混乱先调信息层级;视觉和个性只能增强理解,不能牺牲事实、结构和执行。
## 1. Ground it in the document job
写作或改写前,先判断这篇文档的工作:
- 读者是谁:决策者、执行者、评审者、客户、开发者、运营者,还是未来维护者?
- 读者要完成什么:理解背景、做出决策、执行任务、评审方案、排查问题、学习方法、传播内容,还是留档备查?
- 读完后读者应该能做什么:拍板、执行、确认风险、复用、转发、评论、排查或完成任务?
- 文档寿命是什么:一次性沟通、项目周期内维护、长期知识库、正式归档,还是外部传播?
如果用户没有明确这些信息,能安全假设时说明假设并继续;事实来源、产品口径、结论归属不确定时,先询问或标注待确认,不要编造确定性。
## 2. Route by genre
体裁决定结构、语气、证据密度和表达方式。不要用同一套“背景 / 目标 / 方案 / 风险 / 下一步”应对所有任务。
当用户明确或暗示体裁时,只读取一个最匹配的体裁文件;不要读取整个 `genres/` 目录。简单编辑已有文档时,优先沿用原文体裁和风格,不额外读取体裁文件。
| 用户意图 / 信号词 | 读取文件 |
|---|---|
| 通知、制度、公告、汇报、立项、正式方案、对外说明 | `genres/formal-doc.md` |
| 架构、接口、API、部署、迁移、排障、研发方案、技术设计 | `genres/technical-doc.md` |
| 小红书、爆文、种草、笔记、攻略、社媒传播、标题优化 | `genres/social-xhs.md` |
| PRD、需求说明、用户故事、产品方案、功能设计 | `genres/prd.md` |
| 复盘、事故总结、项目总结、周报、月报、阶段总结 | `genres/retrospective.md` |
| SOP、操作手册、培训材料、入门指南、使用说明、FAQ | `genres/sop-tutorial.md` |
如果任务混合多个体裁,优先选择决定**内容结构**的主文件;语气可以参考用户要求调整,不要为轻微语气差异读取第二个体裁文件。例如“写给领导看的技术方案”优先读 `technical-doc.md`,并采用更正式克制的语气。
## 3. Structure is information
- 结构不是装饰。标题、顺序、列表、表格、callout、grid、画板和引用块都应该表达真实的信息关系。
- 选择结构前先判断:这是顺序、并列、对比、因果、层级、状态、证据,还是行动关系?读者需要先知道什么,才能理解后面的内容?哪些信息应该前置为结论,哪些适合压缩或放到附录?
- 文档开头不是固定的“背景”。决策文档先给结论和待拍板问题;执行文档先给目标、范围和时间线;复盘文档先给事实结论、影响和后续动作;技术文档先给适用范围、前置条件和快速路径;传播文档先给痛点、利益点或钩子。
- 较长、复杂或面向传播 / 汇报 / 决策的文档,最好有一个主组织锚点,如一句话结论、决策矩阵、风险雷达、路线图、架构图、时间线、对比表或 checklist。只保留一个不要到处制造亮点短文档、微编辑、正式归档或简单 SOP 不必强行设计。锚点必须服务事实准确、信息层级和协作执行,不能为了个性化表达增加理解成本。
## 4. Rich blocks must earn their place
飞书富 block 是表达手段,不是装饰指标。不要为了“丰富”“高级”“专业”而加入固定比例的 callout、grid、table 或 whiteboard优先选择最能降低理解成本的表达方式。
| 信息关系 / 场景 | 优先表达 |
|---|---|
| 判断、背景、解释、理由 | 段落 / 小标题 |
| 并列项、短步骤、注意点 | 列表 |
| 待办、检查项、验收项 | checkbox |
| 多对象、多字段、排期、指标、接口参数 | table |
| 少数组并列信息、轻量对比、模块卡片 | grid |
| 风险、限制、例外、强约束、待确认 | callout |
| 外部意见、原话、会议纪要、引用材料 | blockquote |
| 代码、命令、配置、日志 | pre / code |
| 流程、架构、依赖、路线图、因果链路、复杂模型 | whiteboard |
- 组件由信息关系决定不由重要度决定。重要结论可以是段落、callout、表格或画板选择标准是它能否让读者更快理解、比较、判断或行动。
- 关键章节需要判断是否存在图示机会,画板用于图示明显降低理解成本的关系:流程、架构、依赖、时间线、因果、分层、组织关系、复杂对比。结构简单或文字更清楚时,不要强行画板化。具体 Mermaid / SVG / SubAgent 路由见 `lark-doc-whiteboard.md`
- 颜色只在帮助识别语义时使用;正式文档、技术文档、长期知识库和归档材料优先保持朴素。
## 5. Writing style
- 从读者一侧写,而不是从系统实现一侧写。用读者能识别和控制的事物命名,默认使用具体、主动、可执行的表达。
- 具体比聪明更好。一致比花哨更重要。文档词汇是读者导航的路标,同一动作、状态、对象在全文中保持同名。
## 6. Calibrate against AI-doc defaults
写作或改写前,主动避开这些 AI 文档默认值:
1. 不管任务是什么都套“背景 / 目标 / 方案 / 风险 / 下一步”。
2. 用“赋能、闭环、抓手、沉淀、拉通、对齐”等空泛词替代具体动作。
3. 开头写摘要,但没有结论、取舍、风险或行动。
4. 为了“丰富”强行加入 callout、table、grid 或 whiteboard。
5. 把所有重要信息都画板化。
6. 标题整齐,但章节之间没有真实推进关系。
7. 章节之间像多个 Agent 拼接,术语、语气、事实口径不一致。
8. 改写已有文档时全文覆盖,破坏评论、引用、图片、画板或资源块。
如果你的计划看起来像面对任何类似请求都会写出的文档,就修改它,使结构、语气和组件选择更贴合当前读者、任务和体裁。
## 7. Final self-check
交付前检查:
| 指标 | 自检问题 |
|---|---|
| 读者任务 | 这篇文档是否帮助目标读者完成理解、判断、决策或行动? |
| 体裁匹配 | 结构、语气和证据密度是否符合当前体裁? |
| 信息结构 | 标题、顺序、列表、表格和组件是否表达真实关系? |
| 阅读负担 | 是否有段落过长、层级过深、表格过宽或组件过多? |
| 组件必要性 | callout、grid、table、whiteboard 是否真的降低理解成本? |
| 风格一致 | 是否延续用户给定样例或已有文档风格? |
| 事实保真 | 改写时是否保留事实、引用、图片、附件、资源块和关键措辞? |
| 行动清晰 | 读者下一步应该做什么是否明确? |

View File

@@ -0,0 +1,107 @@
# docs history历史版本与回滚
用于查看 Docx 历史版本、按 `history_version_id` 回滚,以及查询回滚任务状态。
## 安全流程
1. 先用分页接口 `+history-list` 找到目标版本的 `history_version_id`
2. 如果用户指定的是 `revision_id`,不要假设它唯一,也不要把 `revision_id` 直接传给 `+history-revert`。先拉一页并在 `entries[]` 中筛选 `revision_id` 相同的候选;如果未匹配到且 `has_more=true`,继续用 `page_token` 翻页;如果已匹配到候选,最多额外再拉一页补齐可能跨页的相邻候选。最终优先根据用户目标时间与 `edit_time` 的接近程度选择最合适的一条,取同一条的 `history_version_id`;如果没有目标时间,或多个候选无法可靠区分,再向用户展示候选版本(`history_version_id``revision_id``edit_time``name/description`)并确认后回滚。
3. 如果用户指定的是某一时刻但没有指定 `revision_id`,按 `entries[].edit_time` 匹配;优先选择不晚于目标时刻的最近一条历史记录,无法明确匹配时先向用户确认候选版本。
4. 再用 `+history-revert --history-version-id <history_version_id>` 发起回滚。默认最多等待 30 秒;如果返回 `status: running`,记录 `task_id`
5.`+history-revert-status` 轮询 `task_id`,直到状态不再是 `running`
6. 回滚完成后,用 `docs +fetch` 读取文档确认内容。
## 按 revision_id 或时间点回滚
当用户说“回滚到 revision_id=42”“恢复到昨天下午 3 点的版本”这类需求时,流程是:
1. 执行 `docs +history-list --doc <doc>` 获取第一页历史记录;`+history-list` 是分页接口,只有 `has_more=true` 且还需要更多候选时才继续传 `--page-token` 翻页。
2. 如果用户给出 `revision_id`:先筛选当前页中 `entries[].revision_id == 用户给出的 revision_id`。如果未命中且 `has_more=true`,继续拉下一页;如果已经命中候选,最多额外再拉一页,补齐同一个 `revision_id` 可能跨页出现的相邻 `history_version_id`。若用户同时给出目标时间,在候选里选择 `edit_time` 与目标时间最接近的一条;若未给目标时间但候选只有一条,可直接使用;若多个候选无法可靠区分,不要自行取第一条,向用户展示候选并确认。
3. 如果用户只给出时间:用 `entries[].edit_time` 匹配,选择目标时刻之前最近的一条;如果用户表达的是“最接近某时刻”,则选择绝对时间差最小的一条。
4. 从最终匹配条目读取 `history_version_id``history_version_id` 对应服务端 `minor_history.version`,这是回滚接口需要的 ID。
5. 执行 `docs +history-revert --doc <doc> --history-version-id <history_version_id>`
候选确认时使用类似格式:
```text
同一个 revision_id 命中多个历史版本,请确认要回滚哪一条:
- history_version_id=11 revision_id=42 edit_time=2026-06-22T12:24:45Z name=...
- history_version_id=12 revision_id=42 edit_time=2026-06-22T12:25:14Z name=...
```
## 命令
```bash
# 列出历史版本
lark-cli docs +history-list --doc "<docx_url_or_token>" --page-size 20
# 翻页
lark-cli docs +history-list --doc "<docx_url_or_token>" --page-size 20 --page-token "<page_token>"
# 回滚到指定 history_version_id默认等待 30000ms
lark-cli docs +history-revert --doc "<docx_url_or_token>" --history-version-id 42
# 只发起任务,不等待
lark-cli docs +history-revert --doc "<docx_url_or_token>" --history-version-id 42 --wait-timeout-ms 0
# 查询回滚任务状态
lark-cli docs +history-revert-status --doc "<docx_url_or_token>" --task-id "<task_id>"
```
## 参数
| 命令 | 参数 | 必填 | 说明 |
|-|-|-|-|
| `+history-list` | `--doc` | 是 | Docx URL/token或可解析为 Docx 的 wiki URL |
| `+history-list` | `--page-size` | 否 | 返回条数,范围 `1-20`,默认 `20` |
| `+history-list` | `--page-token` | 否 | 上一页返回的 `page_token` |
| `+history-revert` | `--doc` | 是 | Docx URL/token或可解析为 Docx 的 wiki URL |
| `+history-revert` | `--history-version-id` | 是 | `+history-list` 返回的 `history_version_id`,必须大于 0 |
| `+history-revert` | `--wait-timeout-ms` | 否 | 等待回滚完成的毫秒数,范围 `0-30000`,默认 `30000` |
| `+history-revert-status` | `--doc` | 是 | 同一个文档 |
| `+history-revert-status` | `--task-id` | 是 | `+history-revert` 返回的 `task_id` |
## 返回值要点
`+history-list` 返回:
```json
{
"entries": [
{
"revision_id": 42,
"history_version_id": "11",
"edit_time": "1780000000",
"type": 1,
"name": "版本名",
"description": "版本说明",
"editor_ids": ["ou_xxx"]
}
],
"has_more": true,
"page_token": "page_token"
}
```
`+history-revert` 返回:
```json
{
"task_id": "task_xxx",
"status": "running",
"history_version_id": "11",
"poll_after_ms": 10000
}
```
`+history-revert-status` 返回:
```json
{
"status": "partial_failed",
"history_version_id": "11",
"failed_block_tokens": ["blk_xxx"]
}
```
`status` 可能是 `running``done``partial_failed``failed`。当状态是 `partial_failed``failed` 时,优先检查 `failed_block_tokens`

View File

@@ -1,11 +1,12 @@
# docs +update更新飞书云文档
> **前置条件MUST READ**
> - 仅做明确旧文本 → 新文本的简单替换、错别字修正、日期 / 人名 / 数字替换时,读取本文件即可;写后必须 fetch 验证。
> - 涉及新增正文、重写段落、调整结构、润色语气、美化排版、补充图表或保真改写时,必须读取:
> 1. [`lark-doc-design-philosophy.md`](lark-doc-design-philosophy.md) — 文档设计判断(读者任务、结构取舍、组件选择、反模板化
> 2. [`lark-doc-xml.md`](lark-doc-xml.md) — XML 语法规则(使用 Markdown 格式时改读 [`lark-doc-md.md`](lark-doc-md.md)
> **前置条件MUST READ** 生成文档内容前,必须先用 Read 工具读取以下文件,缺一不可:
> 1. [`lark-doc-xml.md`](lark-doc-xml.md) — XML 语法规则(使用 Markdown 格式时改读 [`lark-doc-md.md`](lark-doc-md.md)
> 2. [`lark-doc-style.md`](style/lark-doc-style.md) — 写作原则(默认段落、按体裁、组件克制)
> 3. [`lark-doc-update-workflow.md`](style/lark-doc-update-workflow.md) — 改写增强工作流Code-Act Loop、单 Agent 串行改写
>
> **未读完以上文件就生成内容会导致格式错误。**
通过八种指令精确更新飞书云文档。支持字符串级别和 block 级别的操作。
@@ -15,26 +16,6 @@
>
> **Markdown 局限 & block ID 前提:** Markdown 不携带 block ID也无样式颜色、对齐、callout 等)。需要按 block ID 定位(`block_*` 指令的 `--block-id`)时,先 `docs +fetch --detail with-ids` **配合 `--scope``outline` / `range` / `keyword` / `section`)局部获取**目标段落,不要全量 fetch。拿到 block ID 后 `--content` 仍可用 Markdown只是写入内容不带样式。
## Observe-Diagnose-Patch Loop
适用于修改已有飞书文档:调整语气、精简冗余、增补章节、修复结构混乱、按领导意见修改,或在已有图片、引用、表格、评论、资源块的文档上做保真改写。
> [!IMPORTANT]
> 核心原则:先观察,再诊断,再局部 patch最后 fetch 验证。这个流程比全文重写安全;除非用户明确要求完全重建,或文档确实已无保留价值,不要轻易使用 `overwrite`,否则会丢失评论和未支持的资源。
> 每次 `docs +update` 后,都按 block ID 已发生变更处理。如果需要继续修改、重复修改同一处内容,或引用刚插入 / 替换后的内容,必须先重新 `docs +fetch --detail with-ids` 拉取最新内容和 block ID再执行下一轮 patch。
1. **Observe读取现状**:先 `docs +fetch` 读取当前文档状态,并按意图选择最小范围。
- 改某一节或大文档:先 `--scope outline --max-depth 2` 找章节,再 `--scope section --start-block-id <标题id> --detail with-ids`
- 精确跨节区间:用 `--scope range --start-block-id xxx --end-block-id yyy`
- 只有模糊关键词:用 `--scope keyword --keyword xxx --context-before 1 --context-after 1 --detail with-ids`
- 明确整篇重构才读 `--detail with-ids` 全文;只读摘要或确认事实时用更轻的 fetch
2. **Diagnose诊断问题**:判断用户目标、当前结构、语气、重复、断流、事实口径和需要保留的资源;识别哪些 block 必须原样保留。
3. **Patch Plan制定局部计划**:把修改拆成最小安全操作:简单行内替换用 `str_replace`;整段/整块重写用 `block_replace`;增补章节用 `block_insert_after`;删冗余用 `block_delete`;调整顺序用 `block_move_after`
4. **Patch精确修改**:按 block / section 执行局部命令。保护 `<cite>``<img>``<source>``<whiteboard>``<sheet>``<bitable>``<synced_reference>` 等 token 化内容,不要改成纯文本或占位符。同一 block 的多处修改合并成一次 `block_replace`
5. **Verifyfetch 验证)**:每轮写操作后按影响范围重新 fetch检查用户要求、结构、语气、事实、资源块和 block ID 是否符合预期;不满足就基于最新 fetch 结果继续 Diagnose / Patch不要沿用上一轮 block ID。
复杂结构重组时,优先“先插入新结构,再删除旧 block”`block_insert_after` 插入 grid / table / callout / 新章节,再用 `block_delete` 删除旧段落。这样比 `overwrite` 更能保住图片、评论、引用、资源块和不相关内容。`str_replace` 的匹配范围取决于格式XML 模式只适合行内匹配Markdown 模式可跨行和使用 `前缀...后缀`,但跨 block 或容器级重写仍优先用 block 指令。
## 参数
| 参数 | 必填 | 说明 |
@@ -64,12 +45,12 @@
## Block ID 生命周期
安全规则:每次写操作后都按 block ID 已变更处理。需要连续修改、重复修改或操作新插入内容时,必须重新 fetch 最新内容和 block ID不要默认复用之前 fetch 到的 block ID
写操作后不要默认复用之前 fetch 到的 block ID
- `overwrite` / `block_replace` / `block_delete`:受影响旧 ID 失效,继续 block 级操作前必须重新 fetch
- `block_insert_after` / `append` / `block_copy_insert_after`:新内容一定是新 ID要操作新内容或继续编辑插入点附近内容,先重新 fetch
- `block_move_after`位置、章节、range 语义变化;后续依赖位置或章节边界时重新 fetch
- `str_replace`即使是简单行内替换,也不要在后续 block 级操作中假设旧 ID 仍正确;跨行 / 大段替换后必须重新 fetch
- `overwrite` / `block_replace` / `block_delete`:受影响旧 ID 失效,继续 block 级操作前重新 fetch
- `block_insert_after` / `append` / `block_copy_insert_after`锚点 / 源 ID 通常保留,新内容是新 ID要操作新内容先重新 fetch
- `block_move_after`被移动 ID 通常保留,但位置、章节、range 语义变化;后续依赖位置时重新 fetch
- `str_replace`:简单行内替换通常不改变 ID跨行 / 大段替换后如继续 block 级操作,先重新 fetch
## 指令示例
@@ -216,15 +197,63 @@ lark-cli docs +update --doc "<doc_id>" --command block_move_after \
| `warnings` | 警告信息列表 |
| `document.new_blocks` | 本次操作新增的 block 列表(如画板)。`block_id` 可用于后续精确编辑;`block_token` 是资源块 token如画板可交给 `lark-whiteboard` 等 skill 继续操作 |
## 典型工作流
### 精确 block 级更新
1. **获取文档内容和 block ID**
```bash
lark-cli docs +fetch --doc "<doc_id>" --detail with-ids
```
2. **定位目标 block**:从返回的 XML 中找到要修改的 block 及其 `id` 属性
3. **执行更新**
```bash
# 替换特定 block
lark-cli docs +update --doc "<doc_id>" --command block_replace \
--block-id "blkcnXXXX" --content "<p>新内容</p>"
# 在某 block 后插入
lark-cli docs +update --doc "<doc_id>" --command block_insert_after \
--block-id "blkcnXXXX" --content "<h2>追加的章节</h2>"
```
### 简单文本替换
不需要 block ID直接匹配替换
```bash
lark-cli docs +update --doc "<doc_id>" --command str_replace \
--pattern "v1.0" --content "v2.0"
```
## 画板处理
> **`docs +update` 不能直接编辑已有画板的内容。** 本命令只能**新增**画板块;要修改已有画板,先用 `docs +fetch` 取到 `<whiteboard token="...">`,再按 [`lark-doc-whiteboard.md`](lark-doc-whiteboard.md) 启动 SubAgent 读取 [`lark-whiteboard`](../../lark-whiteboard/SKILL.md) 并写入。
新增画板的语法选型见 [`lark-doc-xml.md`](lark-doc-xml.md) 的资源块说明;插入和复杂画板处理见 [`lark-doc-whiteboard.md`](lark-doc-whiteboard.md)。
画板的语法选型与插入示例见 [`lark-doc-xml.md`](lark-doc-xml.md) 与 [`lark-doc-whiteboard.md`](lark-doc-whiteboard.md)。
## 最佳实践
- **精确操作优于全文覆盖**:使用 `block_replace`/`block_insert_after` 精确修改,避免 `overwrite` 全文覆盖
- **str_replace 的匹配范围取决于格式**
- **XML 模式(默认)**`--pattern` 只支持**行内**匹配,不支持跨行 / 跨 block。段落、整块或容器级列表、表格、分栏、引用块等改动请改用 `block_replace` 指定 block_id 重建。
- **Markdown 模式**`--doc-format markdown``--pattern` 同时支持**行内和跨行**匹配,还支持 `前缀...后缀` 省略号语法(用 `...` 串联首尾片段匹配一大段内容),可以一次替换多行文本;但仍建议优先按最小片段匹配,跨 block 容器级重写仍优先用 `block_replace`,避免副作用。
- **保护不可重建的内容**:图片、画板、电子表格等以 token 形式存储,替换时避开这些 block
- **str_replace 的 replacement 支持富文本**:可以用行内标签 `<b>`、`<a>`、`<cite>`、`<latex>` 等替换普通文本为富文本
- **同一 block 只能被 replace 一次**:多次修改同一 block 请合并为一次 block_replace
- **block_delete 支持批量**:用逗号分隔多个 block_id 一次删除
- **复杂结构重组**:将多个段落转换为 grid / table 等复杂布局时,分步操作比 overwrite 更安全:
1. 用 `block_insert_after` 在目标位置插入新的富文本结构
2. 用 `block_delete` 批量删除旧的 block
3. 这样可以保留文档中其他不相关的内容(图片、评论等)
- **表达形式**:插入或替换内容时,优先沿用用户要求和已有文档风格;需要结构化表达时可参考 [`lark-doc-style.md`](style/lark-doc-style.md),但不要为了固定丰富度主动添加组件
## 参考
- [`lark-doc-design-philosophy.md`](lark-doc-design-philosophy.md) — 文档设计判断与组件取舍
- [`lark-doc-update-workflow.md`](style/lark-doc-update-workflow.md) — 改写增强工作流Code-Act Loop、单 Agent 串行改写)
- [`lark-doc-style.md`](style/lark-doc-style.md) — 文档写作原则(默认段落、按体裁、组件克制)
- [`lark-doc-xml.md`](lark-doc-xml.md) — XML 语法规范
- [`lark-doc-fetch.md`](lark-doc-fetch.md) — 获取文档
- [`lark-doc-create.md`](lark-doc-create.md) — 创建文档

View File

@@ -6,7 +6,7 @@
| Skill | 核心职责 | 约束 |
|-------------------|-----------------------------------------------------------|---------------------------------|
| `lark-doc` | 识别画板机会、使用 Mermaid/SVG 创建图表、调度 SubAgent、插入简单图表或复杂空白画板 | 简单图可由主 Agent 直接写入;复杂图再隔离到 SubAgent |
| `lark-doc` | 识别画板机会、使用 Mermaid/SVG 创建图表、调度 SubAgent、插入简单 SVG 画板或复杂空白画板 | 主 Agent 直接创作画板内容; |
| `lark-whiteboard` | 查询/导出已有画板复杂图表生成Mermaid/DSL/SVG 路由、场景选型、渲染验证);写入已有/空白画板 | 仅特别复杂的图表或已有画板更新时由独立 SubAgent 读取 |
## 画板适用规则
@@ -29,9 +29,11 @@
> [!IMPORTANT]
> ⚠️ **分别对每个图表进行决策**
如果有多个位置需要插入图表,你需要根据每个图表的内容**分别决定**采用步骤 2A 还是 2B。思维导图、时序图、类图、饼图、甘特图可插入 mermaid 块;其他类型图表使用 SVG简单图由主 Agent 直接写入,复杂图再启动 SubAgent。
如果有多个位置需要插入图表,你需要根据每个图表的内容**分别决定**采用步骤 2A 还是 2B
中的方式插入这个图表。在需要插入思维导图、时序图、类图、饼图、甘特图的时候可以插入 mermaid 块,在需要插入其他类型图表时启动
SubAgent 插入 SVG。
简单 Mermaid / SVG 图可由主 Agent 直接写入本地 XML需要专门视觉设计、信息密度较高或容易布局翻车的 SVG再启动 SubAgent 产出完整片段。
建议优先使用 SVG 插入图表,除非其属于思维导图、时序图、类图、饼图、甘特图这类可以直接使用 mermaid 语法描述,且不适宜用 SVG 绘制的图表
### 步骤 2A: 使用 mermaid 插入图表
@@ -42,9 +44,9 @@
</whiteboard>
```
### 步骤 2B: 使用 SVG 插入图表
### 步骤 2B: SubAgent 使用 SVG 插入图表
简单 SVG 可由主 Agent 直接写入草稿;复杂 SVG 则启动 SubAgent让它用 `docs +create` / `docs +update` 插入:
主 Agent 启动 SubAgent让它用 `docs +create` / `docs +update` 插入:
```xml
@@ -54,7 +56,7 @@
</whiteboard>
```
复杂 SVG SubAgent 需要携带以下的最小上下文,以及后续的 [SVG 设计 Workflow] 章节指南:
Sub Agent 需要携带以下的最小上下文,以及后续的 [SVG 设计 Workflow] 章节指南:
- doc token、插入位置标题 / block_id / command
- 图表目标、受众、源段落或数据

View File

@@ -13,6 +13,13 @@ lark-cli docs +fetch --doc "$URL" --doc-format xml --detail full --format json \
`$URL` 可以是用户给出的 docx/wiki URL也可以是可被 `docs +fetch` 解析的 token。
## 统计范围
先判断用户要求的是**整篇文档**还是**局部内容**
- 整篇文档的总字数 / 总字符数:按上方「调用方式」抓取 `full` 内容后统计。
- 本次新增 / 替换 / 改写片段的字数:优先统计拟写内容本身;内容已写入文档时,只 fetch 对应 block / range 后统计。不得用整篇文档字数对比局部目标。
如需在自动化或回归验证中发现未覆盖块类型,追加严格参数:
```bash
@@ -36,6 +43,15 @@ lark-cli docs +fetch --doc "$URL" --doc-format xml --detail full --format json \
如果 `unknown_blocks``unsupported_blocks` 非空,回复用户时要说明“已统计可提取文本,但存在未覆盖块,结果可能偏低”,并列出对应块类型。为空时可直接给出结果。
## 字数遵循校验
当用户给了明确字数要求(写 N 字 / x-y 字 / x 字左右 / 上下浮动)时执行;没有明确字数要求则跳过。字数必须按本文流程用脚本统计,不要自己估。
1. 先按「统计范围」确认统计对象,再把要求归一成目标区间:`>x``[x+1, +∞)``<y``(-∞, y-1]``x-y``[x, y]``x 字左右``[round(0.9x), round(1.1x)]`
2. 按统计对象选择对应输入并调用脚本统计实际字数,读取输出里的 `word_count`
3. 对比 `word_count` 与目标区间:区间内即通过;低于下限 → 补充**实质内容**(非注水);高于上限 → 删减冗余内容。改完重新统计
4. **最多 2 轮**。2 轮后仍不达标:停止,不得为达标而注水或删关键内容;如实汇报【目标区间 / 当前字数 / 差值与方向 / 已试 2 轮 / 未达原因】,**禁止谎称达标**
## 输出示例
输入正文等价于:`标题` + `一个苹果是 an apple。` 时,输出形态如下:

View File

@@ -41,7 +41,7 @@ p, h1-h9, ul, ol, li, table, thead, tbody, tr, th, td, blockquote, pre, code, hr
文档中可嵌入外部资源块(属于容器标签的特殊形式),需要额外语法创建:
- `<img>``<img href="https://..."/>` 上传网络图片
- `<whiteboard>` — 简单 Mermaid / SVG 图可由主 Agent 直接写入;复杂 SVG 可启动 SubAgent 产出完整 `<whiteboard type="svg">完整自包含 SVG</whiteboard>`特别复杂或已有画板更新,使用 `<whiteboard type="blank"></whiteboard>` 先创建空白画板,再按 [`lark-doc-whiteboard.md`](lark-doc-whiteboard.md) 启动 SubAgent 调用 `lark-whiteboard` 写入;
- `<whiteboard>` — 简单图由 SubAgent 直接插入 `<whiteboard type="svg">完整自包含 SVG</whiteboard>`复杂图使用 `<whiteboard type="blank"></whiteboard>` 先创建空白画板,再按 [`lark-doc-whiteboard.md`](lark-doc-whiteboard.md) 启动 SubAgent 调用 `lark-whiteboard` 写入;
- `<sheet>``<sheet type="blank"></sheet>` 空白;`<sheet sheet-id="SID" token="TOKEN"></sheet>` 复制已有
- `<task>``<task task-id="GUID"></task>`,必传 task-id任务 guid
- `<chat_card>``<chat_card chat-id="CHAT_ID"></chat_card>`,必传 chat-id

View File

@@ -0,0 +1,47 @@
# 从零创作工作流
用户提供主题、需求或简要说明,需要生成一份新的飞书文档时,遵循本工作流。
## 核心方法论 — Code-Act Loop
通过自适应的 **Code-Act Loop** 驱动文档创作,而非固定模板式的工作流。每次任务都循环执行:
1. **Plan规划** — 根据用户目标和文档当前状态,评估下一步该做什么
2. **Execute执行** — 由主 Agent 自己运行 `lark-cli docs` 命令推进正文;仅画板渲染按需隔离到 SubAgent见步骤三
3. **Observe观察** — 检查命令输出,验证正确性,确认内容是否满足用户目标
4. **Iterate迭代** — 如需调整,回到 Plan 继续循环
循环在文档达到质量标准且满足用户需求时结束。不要试图一次性产出完美内容——迭代打磨效果更好。根据用户实际需求灵活决定文档结构和版块,而不是套用固定模板。
## 典型 Code-Act Loop 流程
### 步骤一:规划与撰写(单 Agent 串行)
正文由主 Agent 串行维护,**不按章节拆给并行 Agent**,避免上下文割裂、重复矛盾和全文级约束失效。
1. 分析用户需求:受众、目的、范围
2. 设计大纲根据任务自然选择结构。可以是短文、纪要、FAQ、方案、报告、清单或其他形式不要默认套固定章节、固定开头或固定富 block 配比
3. `docs +create` 创建并撰写:
- **短文档**:一次写入完整内容。使用 Markdown 时,避免同时传入 `--title` 和同名 `# 标题`
- **长文档**:先建骨架(标题 + 各级标题),再由主 Agent **顺序逐节**用 `block_insert_after --block-id <章节标题 block_id>` 补全正文;写完一节再写下一节,始终带着已写内容的上下文,保证衔接、不重复
- ⚠️ 不要一次性把超长完整内容塞进 `--content`,容易触发字符/参数限制;长文按节分次写入
- ⚠️ 同一节内多次插入时,要锚到**上一个新插入的 block**(按 [`lark-doc-update.md`](../lark-doc-update.md) 的「Block ID 生命周期」),否则反复锚同一个标题会让段落顺序颠倒
- ⚠️ 若先建骨架写了占位摘要,补正文时**删除占位摘要**,不要留残渣
- ⚠️ **`@file` 路径限制**`--content @file` 只接受当前工作目录下的相对路径,传绝对路径(如 `@/tmp/xxx.md`)会报 `unsafe file path`。需要落盘时,将文件写在 cwd 下,用完自行清理
### 步骤二:整合审查与画板识别(串行)
4. `docs +fetch --api-version v2 --detail with-ids` 获取文档,审查整体效果
5. 评估内容是否满足用户目标:事实是否完整、结构是否清楚、语气是否匹配、是否保留必要素材;检查跨节有无重复、矛盾或断流。再按 `lark-doc-style.md` 的「写完自检」快速核对,发现问题就地定向修正
6. **画板识别**:逐章节扫描,判断是否有段落用图明显比文字更易懂(流程 / 架构 / 时间线 / 对比 / 占比等,见 `lark-doc-style.md` 的画板原则。默认用文字只有确需图示才记录需要插图的章节、推荐画板类型、mermaid/SVG 路径和用于画图的源内容
### 步骤三:画板处理与润色
7. **优先处理步骤二识别出的画板需求**:读取并按 [lark-doc-whiteboard.md](../lark-doc-whiteboard.md) 选型和插入;正文本身不交给 SubAgent
8. 由**主 Agent 自行润色**(不另起内容子 Agent正文始终一人维护文字密集且不易读时优先拆段、加小标题或调整顺序——叙述内容保持成段**不要默认改成列表**,只有确属并列要点 / 步骤才用列表(见 `lark-doc-style.md`);只有确实存在行列数据时才用 `<table>`。其余富 block 的取舍一律遵循 `lark-doc-style.md` 的写作原则,不主动堆叠。需要明显分隔的主题可补充 `<hr/>`,不强制章节间都使用。本地图片使用 `docs +media-insert` 插入
### 步骤四:专项校验
9. **字数门禁**如果用户给出任何明确字数要求如“700-800 字”“1000 字左右”“不少于 500 字”“控制在 800 字以内”),本步骤必须执行,不属于按需项。读取并执行 [`lark-doc-word-stat.md`](../lark-doc-word-stat.md) 的「字数遵循校验」;未得到脚本统计结果前,不得向用户声明“符合字数要求”。若没有明确字数要求,则跳过本项,不读取该 workflow。若执行了专项校验向用户呈现目标区间、`word_count` 和达标结论
10. **重复标题检查**:文档生成后,检查文档标题和正文第一个标题块是否重复;若重复,删除或改写正文第一个标题块,避免读者看到同一标题连续出现

View File

@@ -0,0 +1,68 @@
# 飞书文档写作原则
写飞书文档,像一个该领域资深的人类作者那样写,而不是把内容"装配"成组件。
本文只讲"何时用、什么风格";具体标签 / 命令语法见 [`lark-doc-xml.md`](../lark-doc-xml.md)。
## 一、用户明确要求优先
用户点名要某种格式——高亮块、分栏、列表、某编号体例、表格、画板、某模板、某已有文档的风格——**一律照用户的来,下面的"默认克制"全部让位**。用户给了样例或已有文档,就沿用它的结构与语气。
## 二、默认写连贯段落
用户没指定时,**默认是连贯段落**;其余按内容类型分流,别一律"少用结构",也别什么都升标题:
| 内容 | 用什么 | ❌ 别 |
|---|---|---|
| 叙述、论证、分析、说明 | **连贯段落** | 拆成列举 |
| 真·行列数据(预算、指标、对比、排期、字段说明) | **表格** | 写成段落或把字段堆成一行 |
| 字段:值(主题、时长、负责人等,少量) | **加粗标签行**或一句话 | 每字段一个标题 |
| 方法 / 措施 + 每项一段描述 | **加粗引导句段落**(「**全程督导。**…」) | 每项升标题 |
| 任务清单 / 检查项 / 待办事项 | **`<checkbox>`** | 用普通列表替代可交互待办 |
| 纯短并列项(无描述,如材料清单) | 列表 | — |
| 章节(内容成块、需在目录导航) | 标题层级 | — |
- 判断标准:**去掉结构后能顺成段落,就用段落;成行成列的数据,就用表格。**
- **红线一:标题层级只给"章节"。** "小标题 + 一两句话"的小项(字段、方法、要点)不该占标题层级——按上表降成标签行 / 加粗引导句段落(否则目录里全是没信息量的条目)。
- **红线二:列举(「一是 / 二是」「第一 / 第二」「(1)(2)(3)」)只给真正并列的具体项,且别每节都用。**
- 「一是 / 二是」是党务列举的措辞——只用在列具体的**问题 / 措施**那一处;背景、现状、认识、分析、过渡、总结**一律成段**。
- **整篇每段 / 每节都"一是 / 二是",和"每段一个 bullet"是同一个骨架化的错——不因为是党务就变对**(纯清单 / 台账类除外)。
## 三、按体裁写
- **公文 / 法律 / 学术 / 申报 / 项目方案等严肃正式提交物**:靠规范的标题层级、段落与编号体系表达;**默认不用高亮块、分栏**,要强调用加粗或规范小标题。
- **面向公众号、微信等外部平台粘贴 / 发布的内容**:不用飞书特有富 block高亮块、分栏等粘出去会丢样式 / 错乱;改用标准标题、段落、列表、引用。
- **一般文档**:以可读为先,不堆砌结构。
## 四、编号与层级
- **一套编号体例、全篇一致;最忌中文大层级与阿拉伯小数编号混用。**
- 公文 / 正式材料常用:「一、→(一)→ 1.→1中文大层级 + 阿拉伯细分层级)。
- 学术 / 技术 / 商业报告「1 → 1.1 → 1.1.1」或「一、→(一)→ 1.」,**择一**。
- ⚠️ **「一、」只能配「要用阿拉伯小数就从顶层全用「1 / 1.1」。绝不「一、」配「1.1 / 2.1」**——这是最常见的混用。
- **不混用**多套(别"第X部分"+"一、"+"1."混着来);**同级不跳号****不跳级**。
- **编号 / 标题层级只给"章节"**,不要为了凑齐体例把每个小项都编上「(一)」、升成标题(小项处理方式见上文「二、默认写连贯段落」)。
- 简单的 1.2.3 并列项用原生 `<ol><li seq="auto">…</li></ol>` 让飞书自动编号、自动对齐;「一、(一)」原生产不出,才手打成文字——此时用标题级别表达层次,**不靠手动缩进**、各级顶格(全角括号「()」叠手动缩进会视觉错位)。
## 五、飞书特有组件,克制使用
- **高亮块 `<callout>`**:很重的强提醒信号,**默认不用**;只给"不提醒就会出错 / 遗漏"的关键项全文极少0~1 个),不要每节导语 / 结论都做成高亮块。
- **分栏 `<grid>`**:仅左右信息量相当、确需并排对照的短内容;否则用段落或表格。
- **画板**:默认用文字,只在**图示明显比文字更易懂**(流程、架构、时间线、对比、占比等)或用户要求时才用。怎么插、用哪种类型见 [`lark-doc-xml.md`](../lark-doc-xml.md) 与 [`lark-doc-whiteboard.md`](../lark-doc-whiteboard.md)。
- **颜色**:默认朴素、不上色;需要时保持语义一致,按下表选择对应颜色,不为装饰上色。可用色见 [`lark-doc-xml.md`](../lark-doc-xml.md) 的「美化系统」。
| 语义 | 背景色 | 文字色 |
|-|-|-|
| 信息、说明 | `light-blue` | `blue` |
| 成功、推荐 | `light-green` | `green` |
| 警告 / 错误 / 风险 | `light-red` | `red` |
| 注意、待确认 | `light-yellow` | `yellow` |
| 中性、辅助 | `light-gray` | — |
## 六、写完自检
交付前快速回看:
- **叙述是否被列举化**:背景 / 现状 / 认识 / 分析 / 成效 / 过渡 / 总结等应成段;列举只用于同层级、可并列处理的信息,如问题、措施、步骤、任务或材料清单。若正文反复使用连续编号、项目符号或固定并列句式,导致内容缺少叙述,应把背景 / 认识 / 分析 / 过渡改写成有承接关系的段落(纯清单 / 台账类除外)。
- **数据是否正确呈现**:成行成列的数据应使用表格呈现,不要写成段落,也不要用分隔符把多个字段硬串在一起。
- **标题是否滥用**"小标题 + 一句话"的小项不要升成标题;应改成标签行、加粗引导句段落或普通段落。
- **编号是否统一**:全篇一套、不跳号、不跳级,尤其不要中文 + 阿拉伯混用如「一、」配「1.1」)。
- **组件是否克制且保真**:高亮块 / 分栏 / 画板 / 颜色应符合体裁和用户要求;引用 / 图片 / 资源块必须保留。

View File

@@ -0,0 +1,48 @@
# 改写增强工作流
用户提供已有文档链接或 token需要改写、润色、补充或重排版时遵循本工作流。
## 核心方法论 — Code-Act Loop
通过自适应的 **Code-Act Loop** 驱动文档改写,而非固定模板式的工作流。每次任务都循环执行:
1. **Plan规划** — 根据用户目标和文档当前状态,评估下一步该做什么
2. **Execute执行** — 由主 Agent 自己运行 `lark-cli docs` 命令推进改写;仅画板渲染按需隔离到 SubAgent见步骤二
3. **Observe观察** — 检查命令输出,验证正确性,确认内容是否满足用户目标
4. **Iterate迭代** — 如需调整,回到 Plan 继续循环
## 核心原则:精准手术优于全量覆盖
1. **精准手术**:只改用户指定的 block不改其他 block。
2. **全量覆盖**:如果用户明确要改整篇,才用 `overwrite` 命令。
3. **保真约束**:改写时原文里的 `<cite type="user">`@人)、`<cite type="doc">`@文档)、`<img>``<source>``<whiteboard>``<sheet>``<bitable>``<synced_reference>` 等行内组件和资源块一律原样保留(含所有 token / user-id / doc-id 属性),不许替换成纯文本姓名、链接或占位符。
## 工作流程
### 步骤一:分析与画板识别(串行)
1. **选择读取范围**(节省上下文的关键):
- 用户只改某一节 / 文档较大 → 先 `docs +fetch --scope outline --max-depth 2` 拿目录,再 `docs +fetch --scope section --start-block-id <目标标题id> --detail with-ids` 精读该节(`section` 会自动展开到下一个同级/更高级标题前,不用手动算结束 block id
- 需要精确跨节区间 → `docs +fetch --scope range --start-block-id xxx --end-block-id yyy`(或 `--end-block-id -1` 读到末尾)
- 用户只给了模糊关键词 → `docs +fetch --scope keyword --keyword xxx --context-before 1 --context-after 1 --detail with-ids`
- 用户明确要改整篇 → `docs +fetch --detail with-ids`
- 详见 [`lark-doc-fetch.md`](../lark-doc-fetch.md) 中「选 `--scope`(读取范围)」小节
2. 系统性评估:用户想改什么、现有文档风格是什么、哪些内容需要保留、哪些问题影响理解
3. **画板识别**:逐章节扫描,判断是否有段落用图明显比文字更易懂(流程 / 架构 / 时间线 / 对比 / 占比等,见 `lark-doc-style.md` 的画板原则。默认用文字只有确需图示才记录需要插图的章节block ID、推荐画板类型、mermaid/SVG路径和源内容片段
4. 向用户简要说明改进计划(包含识别出的画板机会)
### 步骤二:定向改写(单 Agent 串行)
5. **优先处理步骤一识别出的画板候选段落**:读取并按 [lark-doc-whiteboard.md](../lark-doc-whiteboard.md) 选型和插入;正文本身不交给 SubAgent
6. 由主 Agent **顺序逐节**改写,**不按章节拆给并行 Agent**,避免上下文割裂、重复矛盾和全文级约束失效:
- 沿用或轻微调整已有文档风格,除非用户要求彻底重排版
- 优先通过重写段落、调整标题、补充小标题提升可读性;叙述内容保持成段,**不要默认改成列表**,只有确属并列要点 / 步骤才用列表(见 `lark-doc-style.md`
- 富 block 是可选表达手段,不因固定比例而添加,取舍遵循 `lark-doc-style.md` 的写作原则;画板类需求只走第 5 步
### 步骤三:验证(串行)
7. 获取更新后文档局部内容,检查是否符合用户目标和已有风格
8. 检查是否满足用户目标并保留原有关键内容。再按 `lark-doc-style.md` 的「写完自检」快速核对,发现问题则定向修正
### 步骤四:专项校验(按需执行)
9. 仅当用户预期需要校验字数时,才读取并执行 [`lark-doc-word-stat.md`](../lark-doc-word-stat.md) 的「字数遵循校验」;否则跳过本项,不读取该 workflow。若执行了专项校验向用户呈现结果
**上下文节省提示**:主 Agent 改某节时如需重新读取,优先用 `docs +fetch --scope section --start-block-id <章节标题id>`(自动覆盖整节),或 `--scope range --start-block-id xxx --end-block-id yyy` 精确区间,只拉当前章节,不要重复拉全文。

View File

@@ -29,6 +29,7 @@ metadata:
- 用户要把本地 `.xlsx` / `.csv` / `.base` 导入成 Base / 多维表格 / bitable第一步必须使用 `lark-cli drive +import --type bitable`
- 用户要把本地 `.md` / `.docx` / `.doc` / `.txt` / `.html` 导入成在线文档,使用 `lark-cli drive +import --type docx`
- 用户要把本地 `.pptx` 导入成飞书幻灯片,使用 `lark-cli drive +import --type slides`;当前 PPTX 导入上限是 500MB。
- 批量执行 `drive +import` 且目标是同一个位置(同一 `--folder-token`、默认根目录,或同一 `--target-token`)时,必须串行执行;不要并发导入到同一位置,服务端可能返回并发冲突错误。
- 用户要在 Drive 里上传、创建、读取、局部 patch 或覆盖更新**原生 `.md` 文件**(不是导入成 docx切到 [`lark-markdown`](../lark-markdown/SKILL.md)。
- 用户要比较原生 `.md` 文件的**历史版本差异**,或比较远端 Markdown 与本地草稿,切到 [`lark-markdown`](../lark-markdown/SKILL.md) 的 `lark-cli markdown +diff`;需要版本号时先用 `drive +version-history`
- 用户要查看、下载、回滚或删除文件的**历史版本**,使用 `drive +version-history``drive +version-get``drive +version-revert``drive +version-delete`;这组命令同时支持 `--as user``--as bot`,自动化场景优先 `--as bot`
@@ -102,6 +103,7 @@ lark-cli drive +inspect --url 'https://xxx.feishu.cn/wiki/wikcnXXX'
| `not exist` | 使用了错误的 token | 检查 token 类型wiki 链接必须先查询获取 `obj_token` |
| `permission denied` | 没有相关操作权限 | 引导用户检查当前身份对文档/文件是否有相应操作权限;如果需要,可以授予相应权限 |
| `invalid file_type` | file_type 参数错误 | 根据 `obj_type` 传入正确的 file_typedocx/doc/sheet/slides/bitable |
| `232140101` / `232140100` / `233523001`(常见于 `drive +import``job_error_msg` | 同一位置下存在并发导入 / 创建操作 | 批量导入到同一文件夹、根目录或同一 `--target-token` 时改为串行执行;每个失败项每次重试前等待几秒,总共最多重试 3 次,仍失败就停止并报告冲突 |
### 权限能力入口

View File

@@ -14,6 +14,13 @@
> [!IMPORTANT]
> 当用户**未传 `--name`** 时,文档标题默认取源文件名(去掉扩展名)。在执行导入前,先友好提示用户:「当前未指定文档标题,默认将使用"xxx"作为标题。如果文件内容中也包含相同标题,导入后可能造成视觉重复。是否需要重命名?」让用户确认后再继续。
## 批量导入串行规则
> [!IMPORTANT]
> 批量执行 `drive +import` 且目标是同一个位置时,必须串行执行,不要并发发起导入任务。这里的“相同位置”包括同一个 `--folder-token`、都省略 `--folder-token` 导入到默认根目录,或使用同一个 `--target-token` 导入到已有 bitable。
>
> 如果在同一位置下并发导入,服务端可能返回并发冲突错误。看到错误信息或 `job_error_msg` 中包含 `232140101`、`232140100`、`233523001` 任一错误码时,按同位置并发操作处理:停止并发导入,改为串行处理失败项;每个失败项每次重试前等待几秒,总共最多重试 3 次;仍失败就停止并向用户报告冲突。
## 命令
```bash
@@ -143,6 +150,7 @@ lark-cli drive +import --file ./README.md --type docx --dry-run
- “超过 20MB 自动切换分片上传”只表示上传链路会切到 multipart不代表所有格式都允许导入超过 20MB 的文件。
- 若导入任务执行失败,会返回失败时的 `job_status` 及错误信息。
- 若导入失败信息包含 `232140101``232140100``233523001`,通常表示同一位置下存在并发导入 / 创建操作;批量场景请改为串行执行,每个失败项每次重试前等待几秒,总共最多重试 3 次,仍失败就停止并报告冲突。
- 若内置轮询超时但任务仍在处理中shortcut 会成功返回,并带上:
- `ready=false`
- `timed_out=true`

View File

@@ -146,6 +146,24 @@ lark-cli update
**重要**:始终使用 `lark-cli update` 更新,它会同时更新 CLI 和 AI Skills。
## JSON 输出契约
`--format json`(默认)下,成功与错误的信封结构不同:
成功信封写入 **stdout**(退出码 0
```json
{ "ok": true, "identity": "user", "data": { "guid": "..." }, "meta": { "count": 1 } }
```
错误信封写入 **stderr**(退出码非 0
```json
{ "ok": false, "identity": "user", "error": { "type": "api", "subtype": "...", "code": 99991679, "message": "...", "hint": "..." } }
```
**判断成功必须用 `ok == true`(或进程退出码 0不要用 `code == 0`**:成功信封没有顶层 `code` / `msg` 字段,`code` 只出现在错误信封的 `error` 内,含义是上游 OpenAPI 的 numeric code。按 OpenAPI 老格式 `{"code": 0, "msg": "ok"}` 判断会把所有成功调用误判为失败;封装写入类命令(如 `task +create`)时尤其危险,误判会绕过幂等逻辑导致重复创建。
## 安全规则
- **禁止输出密钥**appSecret、accessToken到终端明文。

View File

@@ -46,7 +46,20 @@ lark-cli task +create --summary "Test Task" --dry-run
1. Confirm with the user: task summary, due date, assignee, and tasklist if necessary.
- **Crucial Rule for Assignee**: If the user explicitly or implicitly says "create a task for me" (给我创建一个任务), or "help me create a task" (帮我新建/创建一个任务), you MUST assign the task to the current logged-in user. You can get the current user's `open_id` by executing `lark-cli auth status` (it already outputs JSON by default, so do not add `--json`) or `lark-cli contact +get-user` first, extracting `.identities.user.openId` (from `auth status`) or `.data.user.open_id` (from `contact +get-user`), and then passing it to the `--assignee` parameter.
2. Execute `lark-cli task +create --summary "..." ...`
3. Report the result: task ID and summary.
3. Judge success by `ok == true` in the stdout JSON (the success envelope has no `code` field — do not test `code == 0`), then report the result: task ID (`data.guid`) and summary.
Example success response:
```json
{
"ok": true,
"identity": "user",
"data": {
"guid": "e297d3d0-4b60-4a5f-a4d4-xxxxxxxxxxxx",
"url": "https://applink.larkoffice.com/client/todo/detail?guid=e297d3d0-4b60-4a5f-a4d4-xxxxxxxxxxxx"
}
}
```
> [!CAUTION]
> This is a **Write Operation** -- You must confirm the user's intent before executing.

View File

@@ -1,9 +1,9 @@
# Docs CLI E2E Coverage
## Metrics
- Denominator: 8 leaf commands
- Covered: 3
- Coverage: 37.5%
- Denominator: 11 leaf commands
- Covered: 6
- Coverage: 54.5%
## Summary
- TestDocs_CreateAndFetchWorkflow: proves `docs +create` and `docs +fetch`; key `t.Run(...)` proof points are `create as bot` and `fetch as bot`.
@@ -11,6 +11,8 @@
- TestDocs_UpdateWorkflow: proves `docs +update` via `update-title-and-content as bot`, then re-fetches the same doc in `verify as bot` to assert persisted title/content changes.
- TestDocs_DryRunDefaultsToV2OpenAPI: proves `docs +create`, `docs +fetch`, and `docs +update` dry-run all emit `/open-apis/docs_ai/v1/...` requests without MCP or `--api-version` guidance; its fetch case asserts fetch sends the default `extra_param`, and its update case asserts `--reference-map` is sent as request body `reference_map`.
- TestDocs_CreateTitleDryRunPrependsContent: proves `docs +create --title` dry-run prepends an escaped `<title>...</title>` tag to request body `content`.
- TestDocs_DryRunDefaultsToV2OpenAPI also proves `docs +history-list`, `docs +history-revert`, and `docs +history-revert-status` dry-run endpoint and query/body shapes.
- TestDocs_HistoryWorkflow proves the guarded live history flow (`LARK_DOC_HISTORY_E2E=1`): create, update, list prior revisions, revert, poll status when needed, and fetch to verify reverted content.
- Setup note: docs workflows create a Drive folder through `drive files create_folder` in `helpers_test.go`; that helper is external to the docs domain and is not counted here.
- Blocked area: media and search shortcuts still need deterministic fixtures and local file orchestration.
@@ -20,6 +22,9 @@
| --- | --- | --- | --- | --- | --- |
| ✓ | docs +create | shortcut | docs/helpers_test.go::createDocWithRetry; docs_create_fetch_test.go::TestDocs_CreateAndFetchWorkflowAsUser/create as user; docs_update_dryrun_test.go::TestDocs_DryRunDefaultsToV2OpenAPI/create; docs_update_dryrun_test.go::TestDocs_CreateTitleDryRunPrependsContent | `--parent-token`; `--doc-format markdown`; `--content`; `--title` | helper asserts returned doc id from `data.document.document_id`; dry-run asserts title is prepended into request body content |
| ✓ | docs +fetch | shortcut | docs_create_fetch_test.go::TestDocs_CreateAndFetchWorkflow/fetch as bot; docs_update_test.go::TestDocs_UpdateWorkflow/verify as bot; docs_create_fetch_test.go::TestDocs_CreateAndFetchWorkflowAsUser/fetch as user; docs_update_dryrun_test.go::TestDocs_DryRunDefaultsToV2OpenAPI/fetch | `--doc <docToken>`; `--doc-format markdown`; default `extra_param.enable_user_cite_reference_map=true` | |
| ✓ | docs +history-list | shortcut | docs_update_dryrun_test.go::TestDocs_DryRunDefaultsToV2OpenAPI/history list; docs_history_workflow_test.go::TestDocs_HistoryWorkflow | `--doc`; `--page-size`; `--page-token` | live workflow gated by `LARK_DOC_HISTORY_E2E=1` |
| ✓ | docs +history-revert | shortcut | docs_update_dryrun_test.go::TestDocs_DryRunDefaultsToV2OpenAPI/history revert; docs_history_workflow_test.go::TestDocs_HistoryWorkflow | `--doc`; `--history-version-id`; `--wait-timeout-ms` | live workflow gated by `LARK_DOC_HISTORY_E2E=1` |
| ✓ | docs +history-revert-status | shortcut | docs_update_dryrun_test.go::TestDocs_DryRunDefaultsToV2OpenAPI/history revert status; docs_history_workflow_test.go::TestDocs_HistoryWorkflow | `--doc`; `--task-id` | live workflow polls only when revert returns `running` |
| ✕ | docs +media-download | shortcut | | none | no media fixture workflow yet |
| ✕ | docs +media-insert | shortcut | | none | requires deterministic upload fixture and rollback assertions |
| ✕ | docs +media-preview | shortcut | | none | requires deterministic media fixture |

View File

@@ -0,0 +1,137 @@
// Copyright (c) 2026 Lark Technologies Pte. Ltd.
// SPDX-License-Identifier: MIT
package docs
import (
"context"
"os"
"testing"
"time"
clie2e "github.com/larksuite/cli/tests/cli_e2e"
"github.com/larksuite/cli/tests/cli_e2e/drive"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/tidwall/gjson"
)
func TestDocs_HistoryWorkflow(t *testing.T) {
if os.Getenv("LARK_DOC_HISTORY_E2E") != "1" {
t.Skip("set LARK_DOC_HISTORY_E2E=1 to run docs history live workflow")
}
clie2e.SkipWithoutUserToken(t)
parentT := t
ctx, cancel := context.WithTimeout(context.Background(), 3*time.Minute)
t.Cleanup(cancel)
suffix := clie2e.GenerateSuffix()
folderName := "lark-cli-e2e-docs-history-folder-" + suffix
docTitle := "lark-cli-e2e-docs-history-" + suffix
originalMarker := "original history marker " + suffix
updatedMarker := "updated history marker " + suffix
const defaultAs = "user"
folderToken := drive.CreateDriveFolder(t, parentT, ctx, folderName, defaultAs, "")
docToken := createDocWithRetry(t, parentT, ctx, folderToken, docTitle, originalMarker, defaultAs)
updateResult, err := clie2e.RunCmd(ctx, clie2e.Request{
Args: []string{
"docs", "+update",
"--doc", docToken,
"--command", "overwrite",
"--doc-format", "markdown",
"--content", "# " + docTitle + "\n\n" + updatedMarker,
},
DefaultAs: defaultAs,
})
require.NoError(t, err)
updateResult.AssertExitCode(t, 0)
updateResult.AssertStdoutStatus(t, true)
fetchUpdated, err := clie2e.RunCmd(ctx, clie2e.Request{
Args: []string{"docs", "+fetch", "--doc", docToken, "--doc-format", "markdown"},
DefaultAs: defaultAs,
})
require.NoError(t, err)
fetchUpdated.AssertExitCode(t, 0)
fetchUpdated.AssertStdoutStatus(t, true)
updatedContent := gjson.Get(fetchUpdated.Stdout, "data.document.content").String()
assert.Contains(t, updatedContent, updatedMarker)
currentRevision := gjson.Get(fetchUpdated.Stdout, "data.document.revision_id").Int()
require.Greater(t, currentRevision, int64(0), "stdout:\n%s", fetchUpdated.Stdout)
var revertHistoryVersionID string
require.Eventually(t, func() bool {
listResult, listErr := clie2e.RunCmd(ctx, clie2e.Request{
Args: []string{
"docs", "+history-list",
"--doc", docToken,
"--page-size", "20",
},
DefaultAs: defaultAs,
})
require.NoError(t, listErr)
listResult.AssertExitCode(t, 0)
listResult.AssertStdoutStatus(t, true)
for _, entry := range gjson.Get(listResult.Stdout, "data.entries").Array() {
revisionID := entry.Get("revision_id").Int()
historyVersionID := entry.Get("history_version_id").String()
if revisionID > 0 && revisionID < currentRevision && historyVersionID != "" {
revertHistoryVersionID = historyVersionID
return true
}
}
return false
}, 45*time.Second, 3*time.Second, "history list did not expose a prior revision")
require.NotEmpty(t, revertHistoryVersionID)
revertResult, err := clie2e.RunCmd(ctx, clie2e.Request{
Args: []string{
"docs", "+history-revert",
"--doc", docToken,
"--history-version-id", revertHistoryVersionID,
"--wait-timeout-ms", "30000",
},
DefaultAs: defaultAs,
})
require.NoError(t, err)
revertResult.AssertExitCode(t, 0)
revertResult.AssertStdoutStatus(t, true)
status := gjson.Get(revertResult.Stdout, "data.status").String()
taskID := gjson.Get(revertResult.Stdout, "data.task_id").String()
statusStdout := revertResult.Stdout
if status == "running" {
require.NotEmpty(t, taskID, "stdout:\n%s", revertResult.Stdout)
require.Eventually(t, func() bool {
statusResult, statusErr := clie2e.RunCmd(ctx, clie2e.Request{
Args: []string{
"docs", "+history-revert-status",
"--doc", docToken,
"--task-id", taskID,
},
DefaultAs: defaultAs,
})
require.NoError(t, statusErr)
statusResult.AssertExitCode(t, 0)
statusResult.AssertStdoutStatus(t, true)
statusStdout = statusResult.Stdout
status = gjson.Get(statusResult.Stdout, "data.status").String()
return status != "" && status != "running"
}, 60*time.Second, 5*time.Second, "history revert task did not finish")
}
require.Equal(t, "done", status, "status stdout:\n%s", statusStdout)
fetchReverted, err := clie2e.RunCmd(ctx, clie2e.Request{
Args: []string{"docs", "+fetch", "--doc", docToken, "--doc-format", "markdown"},
DefaultAs: defaultAs,
})
require.NoError(t, err)
fetchReverted.AssertExitCode(t, 0)
fetchReverted.AssertStdoutStatus(t, true)
revertedContent := gjson.Get(fetchReverted.Stdout, "data.document.content").String()
assert.Contains(t, revertedContent, originalMarker)
assert.NotContains(t, revertedContent, updatedMarker)
}

View File

@@ -26,7 +26,10 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
tests := []struct {
name string
args []string
wantContains []string
wantURL string
wantParams map[string]any
wantBody map[string]any
wantExtraParam string
wantRefLabel string
}{
@@ -37,7 +40,7 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
"--content", "<title>Dry Run</title><p>hello</p>",
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents",
wantContains: []string{"/open-apis/docs_ai/v1/documents"},
},
{
name: "create api-version v1 compatibility",
@@ -47,7 +50,7 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
"--content", "<title>Dry Run</title><p>hello</p>",
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents",
wantContains: []string{"/open-apis/docs_ai/v1/documents"},
},
{
name: "fetch",
@@ -56,7 +59,7 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
"--doc", "doxcnDryRunE2E",
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents/doxcnDryRunE2E/fetch",
wantContains: []string{"/open-apis/docs_ai/v1/documents/doxcnDryRunE2E/fetch"},
wantExtraParam: `{"enable_user_cite_reference_map":true,"return_html5_block_data":true}`,
},
{
@@ -68,7 +71,7 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
"--content", "<p>hello</p>",
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents/doxcnDryRunE2E",
wantContains: []string{"/open-apis/docs_ai/v1/documents/doxcnDryRunE2E"},
},
{
name: "update reference-map",
@@ -80,7 +83,7 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
"--reference-map", `{"widget":{"r1":{"label":"widget-ref-value"}}}`,
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents/doxcnDryRunE2E",
wantContains: []string{"/open-apis/docs_ai/v1/documents/doxcnDryRunE2E"},
wantRefLabel: "widget-ref-value",
},
{
@@ -92,7 +95,50 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
"--block-id", "blkA,blkB,blkC",
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents/doxcnDryRunE2E",
wantContains: []string{"/open-apis/docs_ai/v1/documents/doxcnDryRunE2E"},
},
{
name: "history list",
args: []string{
"docs", "+history-list",
"--doc", "doxcnDryRunE2E",
"--page-size", "5",
"--page-token", "page_token_1",
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents/doxcnDryRunE2E/histories",
wantParams: map[string]any{
"page_size": 5,
"page_token": "page_token_1",
},
},
{
name: "history revert",
args: []string{
"docs", "+history-revert",
"--doc", "doxcnDryRunE2E",
"--history-version-id", "42",
"--wait-timeout-ms", "0",
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents/doxcnDryRunE2E/history/revert",
wantBody: map[string]any{
"history_version_id": "42",
"wait_timeout_ms": 0,
},
},
{
name: "history revert status",
args: []string{
"docs", "+history-revert-status",
"--doc", "doxcnDryRunE2E",
"--task-id", "task_1",
"--dry-run",
},
wantURL: "/open-apis/docs_ai/v1/documents/doxcnDryRunE2E/history/revert_status",
wantParams: map[string]any{
"task_id": "task_1",
},
},
}
@@ -106,10 +152,7 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
result.AssertExitCode(t, 0)
combined := result.Stdout + "\n" + result.Stderr
for _, want := range []string{
tt.wantURL,
"docs_ai/v1",
} {
for _, want := range append(tt.wantContains, "docs_ai/v1") {
if !strings.Contains(combined, want) {
t.Fatalf("dry-run output missing %q\nstdout:\n%s\nstderr:\n%s", want, result.Stdout, result.Stderr)
}
@@ -120,6 +163,15 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
if strings.Contains(combined, "--api-version") {
t.Fatalf("dry-run output should not ask for --api-version\nstdout:\n%s\nstderr:\n%s", result.Stdout, result.Stderr)
}
if tt.wantURL != "" {
require.Equal(t, tt.wantURL, gjson.Get(result.Stdout, "api.0.url").String(), "stdout:\n%s", result.Stdout)
}
for key, want := range tt.wantParams {
assertDryRunField(t, result.Stdout, "api.0.params."+key, want)
}
for key, want := range tt.wantBody {
assertDryRunField(t, result.Stdout, "api.0.body."+key, want)
}
if tt.wantExtraParam != "" {
extraParam := gjson.Get(result.Stdout, "api.0.body.extra_param").String()
require.JSONEq(t, tt.wantExtraParam, extraParam, "stdout:\n%s", result.Stdout)
@@ -132,6 +184,21 @@ func TestDocs_DryRunDefaultsToV2OpenAPI(t *testing.T) {
}
}
func assertDryRunField(t *testing.T, stdout, path string, want any) {
t.Helper()
got := gjson.Get(stdout, path)
require.True(t, got.Exists(), "%s missing in stdout:\n%s", path, stdout)
switch want := want.(type) {
case int:
require.Equal(t, int64(want), got.Int(), "%s in stdout:\n%s", path, stdout)
case string:
require.Equal(t, want, got.String(), "%s in stdout:\n%s", path, stdout)
default:
t.Fatalf("unsupported dry-run assertion type %T for %s", want, path)
}
}
func TestDocs_CreateTitleDryRunPrependsContent(t *testing.T) {
// Fake creds are enough — dry-run short-circuits before any real API call.
t.Setenv("LARKSUITE_CLI_APP_ID", "app")