fix(skills): align skill guidance with the typed error contract

Skill references written before the typed-error refactor still taught retired envelope shapes. AI agents following them now read what the CLI actually emits: - permission recovery reads error.missing_scopes instead of the upstream permission_violations detail - confirmation gates use type=confirmation, subtype=confirmation_required, and flat risk/action fields - drive duplicate-remote failures are typed validation envelopes (failed_precondition with params[]), not duplicate_remote_path with error.detail - drive batch partial failures are ok:false results on stdout, not an error.type=partial_failure stderr envelope - minutes edit-permission and word-replace misses branch on stable error.code/error.subtype, not message text - slides replace failures are stderr typed envelopes only; no raw backend response is printed to stdout - slides and Base command outputs show the ok/identity/data success envelope instead of the raw {code,msg} OpenAPI wrapper
e2e: harden CLI E2E retry, cleanup, and domain selection (#1709 )
2026-07-08 02:00:19 +08:00 · 2026-07-07 21:45:37 +08:00 · 2026-07-07 19:41:11 +08:00
37 changed files with 1208 additions and 202 deletions
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -263,13 +263,19 @@ jobs:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@93cb6efe18208431cddfb8368fd83d5badbf9bfd # v5
+        with:
+          fetch-depth: 0
      - uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6
        with:
          go-version-file: go.mod
      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6
        with:
          python-version: '3.x'
+      - name: Resolve CLI E2E domains
+        id: e2e_domains
+        run: node scripts/e2e_domains.js
      - name: Build lark-cli
+        if: ${{ steps.e2e_domains.outputs.mode != 'skip' }}
        run: make build
      - name: Run dry-run E2E tests
        env:
@@ -277,7 +283,28 @@ jobs:
          LARKSUITE_CLI_APP_ID: dry-run
          LARKSUITE_CLI_APP_SECRET: dry-run
          LARKSUITE_CLI_BRAND: feishu
-        run: go test -v -count=1 -timeout=5m ./tests/cli_e2e/... -run 'DryRun|Regression'
+          E2E_MODE: ${{ steps.e2e_domains.outputs.mode }}
+          E2E_REASON: ${{ steps.e2e_domains.outputs.reason }}
+          E2E_DRY_ROOT_PACKAGE: ${{ steps.e2e_domains.outputs.dry_root_package }}
+          E2E_DRY_PACKAGES: ${{ steps.e2e_domains.outputs.dry_packages }}
+        run: |
+          if [ "$E2E_MODE" = "skip" ]; then
+            echo "No dry-run CLI E2E needed: $E2E_REASON"
+            exit 0
+          fi
+          if [ -z "$E2E_DRY_ROOT_PACKAGE" ] && [ -z "$E2E_DRY_PACKAGES" ]; then
+            echo "::error::No dry-run CLI E2E packages resolved for mode $E2E_MODE"
+            exit 1
+          fi
+          echo "Dry-run CLI E2E domains: $E2E_MODE ($E2E_REASON)"
+          if [ -n "$E2E_DRY_ROOT_PACKAGE" ]; then
+            echo "Dry-run CLI E2E root package: $E2E_DRY_ROOT_PACKAGE"
+            go test -v -count=1 -timeout=5m "$E2E_DRY_ROOT_PACKAGE"
+          fi
+          if [ -n "$E2E_DRY_PACKAGES" ]; then
+            echo "Dry-run CLI E2E packages: $E2E_DRY_PACKAGES"
+            go test -v -count=1 -timeout=5m $E2E_DRY_PACKAGES -run 'DryRun|Regression'
+          fi

  e2e-live:
    needs: [unit-test, lint, script-test, deterministic-gate]
@@ -292,15 +319,22 @@ jobs:
      TEST_USER_ACCESS_TOKEN: ${{ secrets.TEST_USER_ACCESS_TOKEN }}
    steps:
      - uses: actions/checkout@93cb6efe18208431cddfb8368fd83d5badbf9bfd # v5
+        with:
+          fetch-depth: 0
      - uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6
        with:
          go-version-file: go.mod
      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6
        with:
          python-version: '3.x'
+      - name: Resolve CLI E2E domains
+        id: e2e_domains
+        run: node scripts/e2e_domains.js
      - name: Build lark-cli
+        if: ${{ steps.e2e_domains.outputs.mode != 'skip' }}
        run: make build
      - name: Configure bot credentials
+        if: ${{ steps.e2e_domains.outputs.mode != 'skip' }}
        run: |
          if [ -z "$TEST_BOT1_APP_ID" ] || [ -z "$TEST_BOT1_APP_SECRET" ]; then
            echo "::error::Missing required secrets: TEST_BOT1_APP_ID / TEST_BOT1_APP_SECRET"
@@ -310,16 +344,24 @@ jobs:
      - name: Run CLI E2E tests
        env:
          LARK_CLI_BIN: ${{ github.workspace }}/lark-cli
+          E2E_MODE: ${{ steps.e2e_domains.outputs.mode }}
+          E2E_REASON: ${{ steps.e2e_domains.outputs.reason }}
+          E2E_LIVE_PACKAGES: ${{ steps.e2e_domains.outputs.live_packages }}
        run: |
-          packages=$(go list ./tests/cli_e2e/... | grep -v '^github.com/larksuite/cli/tests/cli_e2e$' | grep -v '/demo$')
+          if [ "$E2E_MODE" = "skip" ]; then
+            echo "No live CLI E2E needed: $E2E_REASON"
+            exit 0
+          fi
+          packages="$E2E_LIVE_PACKAGES"
          if [ -z "$packages" ]; then
-            echo "No CLI E2E packages to test after exclusions."
+            echo "::error::No live CLI E2E packages resolved for mode $E2E_MODE"
            exit 1
          fi
-          packages_arg=$(printf '%s\n' "$packages" | paste -sd' ' -)
-          go run gotest.tools/gotestsum@v1.12.3 --rerun-fails=2 --rerun-fails-max-failures=20 --packages="$packages_arg" --format testname --junitfile cli-e2e-report.xml -- -count=1 -v
+          echo "Live CLI E2E domains: $E2E_MODE ($E2E_REASON)"
+          echo "Live CLI E2E packages: $packages"
+          go run gotest.tools/gotestsum@v1.12.3 --rerun-fails=2 --rerun-fails-max-failures=20 --packages="$packages" --format testname --junitfile cli-e2e-report.xml -- -count=1 -v
      - name: Publish CLI E2E test report
-        if: ${{ !cancelled() }}
+        if: ${{ !cancelled() && steps.e2e_domains.outputs.mode != 'skip' }}
        uses: dorny/test-reporter@a43b3a5f7366b97d083190328d2c652e1a8b6aa2 # v3.0.0
        with:
          name: CLI E2E Tests
--- a/2
+++ b/2
@@ -51,7 +51,7 @@ script-test:
 	bash scripts/resolve-changed-from.test.sh
 	bash scripts/ci-workflow.test.sh
 	bash scripts/semantic-review-workflow.test.sh
-	$(NODE) --test scripts/semantic-review-verify-artifact.test.js scripts/pr-quality-summary.test.js scripts/semantic-review-publish.test.js scripts/ci-quality-summary-publish.test.js
+	$(NODE) --test scripts/e2e_domains.test.js scripts/semantic-review-verify-artifact.test.js scripts/pr-quality-summary.test.js scripts/semantic-review-publish.test.js scripts/ci-quality-summary-publish.test.js

 # ./extension/... keeps the public plugin SDK in the default test matrix.
 unit-test: fetch_meta
--- a/cmd/root_integration_test.go
+++ b/cmd/root_integration_test.go
@@ -14,11 +14,13 @@ import (
 	"github.com/larksuite/cli/cmd/api"
 	"github.com/larksuite/cli/cmd/auth"
 	"github.com/larksuite/cli/cmd/service"
+	"github.com/larksuite/cli/internal/apicatalog"
 	"github.com/larksuite/cli/internal/build"
 	"github.com/larksuite/cli/internal/cmdutil"
 	"github.com/larksuite/cli/internal/core"
 	"github.com/larksuite/cli/internal/envvars"
 	"github.com/larksuite/cli/internal/httpmock"
+	"github.com/larksuite/cli/internal/meta"
 	"github.com/larksuite/cli/internal/output"
 	"github.com/larksuite/cli/internal/skillscheck"
 	"github.com/larksuite/cli/internal/update"
@@ -103,6 +105,11 @@ func parseTypedEnvelope(t *testing.T, stderr *bytes.Buffer) typedErrorEnvelope {
 }

 func buildStrictModeIntegrationRootCmd(t *testing.T, f *cmdutil.Factory) *cobra.Command {
+	t.Helper()
+	return buildStrictModeIntegrationRootCmdWithCatalog(t, f, nil)
+}
+
+func buildStrictModeIntegrationRootCmdWithCatalog(t *testing.T, f *cmdutil.Factory, catalog *apicatalog.Catalog) *cobra.Command {
 	t.Helper()
 	rootCmd := &cobra.Command{Use: "lark-cli"}
 	rootCmd.SilenceErrors = true
@@ -113,7 +120,11 @@ func buildStrictModeIntegrationRootCmd(t *testing.T, f *cmdutil.Factory) *cobra.
 	}
 	rootCmd.AddCommand(auth.NewCmdAuth(f))
 	rootCmd.AddCommand(api.NewCmdApi(f, nil))
-	service.RegisterServiceCommands(rootCmd, f)
+	if catalog != nil {
+		service.RegisterServiceCommandsFromCatalog(context.Background(), rootCmd, f, *catalog)
+	} else {
+		service.RegisterServiceCommands(rootCmd, f)
+	}
 	shortcuts.RegisterShortcuts(rootCmd, f)
 	if mode := f.ResolveStrictMode(context.Background()); mode.IsActive() {
 		pruneForStrictMode(rootCmd, mode)
@@ -121,6 +132,29 @@ func buildStrictModeIntegrationRootCmd(t *testing.T, f *cmdutil.Factory) *cobra.
 	return rootCmd
 }

+func strictModeFixtureCatalog() apicatalog.Catalog {
+	return apicatalog.New(apicatalog.SourceEmbedded, []meta.Service{
+		{
+			Name:        "fixture",
+			ServicePath: "/open-apis/fixture/v1",
+			Resources: map[string]meta.Resource{
+				"things": {
+					Methods: map[string]meta.Method{
+						"create": {
+							Path:         "things",
+							HTTPMethod:   "POST",
+							AccessTokens: []meta.Token{meta.TokenTenant},
+							RequestBody: map[string]meta.Field{
+								"name": {Type: "string"},
+							},
+						},
+					},
+				},
+			},
+		},
+	})
+}
+
 func newStrictModeDefaultFactory(t *testing.T, profile string, mode core.StrictMode) (*cmdutil.Factory, *bytes.Buffer, *bytes.Buffer) {
 	t.Helper()
 	t.Setenv(envvars.CliAppID, "")
@@ -355,10 +389,11 @@ func TestIntegration_StrictModeBot_ProfileOverride_ServiceExplicitUserReturnsEnv

 func TestIntegration_StrictModeUser_ProfileOverride_ServiceBotOnlyMethodReturnsEnvelope(t *testing.T) {
 	f, stdout, stderr := newStrictModeDefaultFactory(t, "target", core.StrictModeUser)
-	rootCmd := buildStrictModeIntegrationRootCmd(t, f)
+	catalog := strictModeFixtureCatalog()
+	rootCmd := buildStrictModeIntegrationRootCmdWithCatalog(t, f, &catalog)

 	code := executeRootIntegration(t, f, rootCmd, []string{
-		"im", "images", "create", "--data", `{"image_type":"message","image":"x"}`, "--dry-run",
+		"fixture", "things", "create", "--data", `{"name":"probe"}`, "--dry-run",
 	})

 	if code != output.ExitValidation {
--- a/scripts/ci-workflow.test.sh
+++ b/scripts/ci-workflow.test.sh
@@ -215,6 +215,73 @@ if ! grep -Fq "if: \${{ $fork_safe_guard }}" <<<"$section"; then
  exit 1
 fi

+if ! grep -Fq "name: Resolve CLI E2E domains" <<<"$dry_run_section" ||
+   ! grep -Fq "id: e2e_domains" <<<"$dry_run_section" ||
+   ! grep -Fq "run: node scripts/e2e_domains.js" <<<"$dry_run_section"; then
+  echo "e2e-dry-run should resolve changed-file CLI E2E domains before running tests"
+  exit 1
+fi
+
+if ! grep -Fq "steps.e2e_domains.outputs.dry_packages" <<<"$dry_run_section"; then
+  echo "e2e-dry-run should use resolved dry_packages instead of always running the full suite"
+  exit 1
+fi
+
+if ! grep -Fq "E2E_REASON: \${{ steps.e2e_domains.outputs.reason }}" <<<"$dry_run_section" ||
+   ! grep -Fq 'echo "Dry-run CLI E2E domains: $E2E_MODE ($E2E_REASON)"' <<<"$dry_run_section"; then
+  echo "e2e-dry-run should pass dynamic domain output through env before shell use"
+  exit 1
+fi
+
+if ! grep -Fq "E2E_DRY_ROOT_PACKAGE: \${{ steps.e2e_domains.outputs.dry_root_package }}" <<<"$dry_run_section" ||
+   ! grep -Fq 'go test -v -count=1 -timeout=5m "$E2E_DRY_ROOT_PACKAGE"' <<<"$dry_run_section"; then
+  echo "e2e-dry-run should run the root CLI E2E harness package without the DryRun/Regression filter"
+  exit 1
+fi
+
+if ! grep -Fq "No dry-run CLI E2E needed" <<<"$dry_run_section"; then
+  echo "e2e-dry-run should explicitly skip when domain mode is skip"
+  exit 1
+fi
+
+if ! grep -Fq "name: Resolve CLI E2E domains" <<<"$section" ||
+   ! grep -Fq "id: e2e_domains" <<<"$section" ||
+   ! grep -Fq "run: node scripts/e2e_domains.js" <<<"$section"; then
+  echo "e2e-live should resolve changed-file CLI E2E domains before credentials and tests"
+  exit 1
+fi
+
+if ! grep -Fq "steps.e2e_domains.outputs.live_packages" <<<"$section"; then
+  echo "e2e-live should use resolved live_packages instead of always running the full suite"
+  exit 1
+fi
+
+if ! grep -Fq "E2E_REASON: \${{ steps.e2e_domains.outputs.reason }}" <<<"$section" ||
+   ! grep -Fq 'echo "Live CLI E2E domains: $E2E_MODE ($E2E_REASON)"' <<<"$section"; then
+  echo "e2e-live should pass dynamic domain output through env before shell use"
+  exit 1
+fi
+
+if ! awk '
+  /^      - name: Build lark-cli/ { in_step = 1 }
+  in_step && /if: \$\{\{ steps\.e2e_domains\.outputs\.mode != '\''skip'\'' \}\}/ { found = 1 }
+  in_step && /^      - name:/ && !/Build lark-cli/ { in_step = 0 }
+  END { exit found ? 0 : 1 }
+' <<<"$dry_run_section"; then
+  echo "e2e-dry-run should skip building lark-cli when domain mode is skip"
+  exit 1
+fi
+
+if ! awk '
+  /^      - name: Build lark-cli/ { in_step = 1 }
+  in_step && /if: \$\{\{ steps\.e2e_domains\.outputs\.mode != '\''skip'\'' \}\}/ { found = 1 }
+  in_step && /^      - name:/ && !/Build lark-cli/ { in_step = 0 }
+  END { exit found ? 0 : 1 }
+' <<<"$section"; then
+  echo "e2e-live should skip building lark-cli when domain mode is skip"
+  exit 1
+fi
+
 if ! grep -Fq "permissions:" <<<"$section" ||
   ! grep -Fq "contents: read" <<<"$section" ||
   ! grep -Fq "checks: write" <<<"$section"; then
@@ -237,13 +304,23 @@ if ! grep -Fq "::error::Missing required secrets: TEST_BOT1_APP_ID / TEST_BOT1_A
  exit 1
 fi

+if ! awk '
+  /^      - name: Configure bot credentials/ { in_step = 1 }
+  in_step && /if: \$\{\{ steps\.e2e_domains\.outputs\.mode != '\''skip'\'' \}\}/ { found = 1 }
+  in_step && /^      - name:/ && !/Configure bot credentials/ { in_step = 0 }
+  END { exit found ? 0 : 1 }
+' <<<"$section"; then
+  echo "e2e-live should only configure bot credentials when domain mode is not skip"
+  exit 1
+fi
+
 if grep -Fq "steps.live_e2e_credentials.outputs.configured" <<<"$section"; then
  echo "e2e-live build, configure, test, and report steps should not be gated by a skip-state output"
  exit 1
 fi

-if ! grep -Fq "if: \${{ !cancelled() }}" <<<"$section"; then
-  echo "e2e-live report step should run after attempted live tests unless the workflow is cancelled"
+if ! grep -Fq "if: \${{ !cancelled() && steps.e2e_domains.outputs.mode != 'skip' }}" <<<"$section"; then
+  echo "e2e-live report step should run after attempted live tests unless the workflow is cancelled or domain mode is skip"
  exit 1
 fi

--- a/scripts/domain-map.js
+++ b/scripts/domain-map.js
@@ -0,0 +1,54 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+const fs = require("node:fs");
+const path = require("node:path");
+
+const DOMAIN_MAP_PATH = path.join(__dirname, "domain-map.json");
+const domainMap = JSON.parse(fs.readFileSync(DOMAIN_MAP_PATH, "utf8"));
+
+function normalizeRepoPath(input) {
+  return String(input || "").trim().replace(/\\/g, "/").replace(/^\.\//, "").toLowerCase();
+}
+
+const pathMappingsBySpecificity = (domainMap.pathMappings || [])
+  .map((entry) => ({ ...entry, prefix: normalizeRepoPath(entry.prefix) }))
+  .sort((a, b) => b.prefix.length - a.prefix.length);
+
+function findPathMapping(filePath) {
+  const normalized = normalizeRepoPath(filePath);
+  return pathMappingsBySpecificity.find((entry) => normalized.startsWith(entry.prefix));
+}
+
+function labelDomainsForPath(filePath) {
+  const mapping = findPathMapping(filePath);
+  return mapping ? [...(mapping.labelDomains || [])] : [];
+}
+
+function e2eDomainsForPath(filePath) {
+  const mapping = findPathMapping(filePath);
+  return mapping ? [...(mapping.e2eDomains || [])] : [];
+}
+
+function matchesFullFallback(filePath) {
+  const normalized = normalizeRepoPath(filePath);
+  return (domainMap.fullFallbackPrefixes || []).some((prefix) => normalized.startsWith(prefix));
+}
+
+function isSkippablePath(filePath) {
+  const normalized = normalizeRepoPath(filePath);
+  const basename = path.posix.basename(normalized);
+  return (domainMap.skipPrefixes || []).some((prefix) => normalized.startsWith(prefix))
+    || (domainMap.skipSuffixes || []).some((suffix) => normalized.endsWith(suffix))
+    || (domainMap.skipFilenames || []).includes(basename);
+}
+
+module.exports = {
+  domainMap,
+  e2eDomainsForPath,
+  findPathMapping,
+  isSkippablePath,
+  labelDomainsForPath,
+  matchesFullFallback,
+  normalizeRepoPath,
+};
--- a/scripts/domain-map.json
+++ b/scripts/domain-map.json
@@ -0,0 +1,71 @@
+{
+  "pathMappings": [
+    { "prefix": "shortcuts/im/", "labelDomains": ["im"], "e2eDomains": ["im"] },
+    { "prefix": "shortcuts/vc/", "labelDomains": ["vc"], "e2eDomains": ["vc"] },
+    { "prefix": "shortcuts/calendar/", "labelDomains": ["calendar"], "e2eDomains": ["calendar"] },
+    { "prefix": "shortcuts/doc/", "labelDomains": ["ccm"], "e2eDomains": ["docs"] },
+    { "prefix": "shortcuts/sheets/", "labelDomains": ["ccm"], "e2eDomains": ["sheets"] },
+    { "prefix": "shortcuts/drive/", "labelDomains": ["ccm"], "e2eDomains": ["drive"] },
+    { "prefix": "shortcuts/wiki/", "labelDomains": ["ccm"], "e2eDomains": ["wiki"] },
+    { "prefix": "shortcuts/base/", "labelDomains": ["base"], "e2eDomains": ["base"] },
+    { "prefix": "shortcuts/mail/", "labelDomains": ["mail"], "e2eDomains": ["mail"] },
+    { "prefix": "shortcuts/task/", "labelDomains": ["task"], "e2eDomains": ["task"] },
+    { "prefix": "shortcuts/contact/", "labelDomains": ["contact"], "e2eDomains": ["contact"] },
+    { "prefix": "shortcuts/apps/", "labelDomains": [], "e2eDomains": ["apps"] },
+    { "prefix": "shortcuts/markdown/", "labelDomains": [], "e2eDomains": ["markdown"] },
+    { "prefix": "shortcuts/minutes/", "labelDomains": [], "e2eDomains": ["minutes"] },
+    { "prefix": "shortcuts/okr/", "labelDomains": [], "e2eDomains": ["okr"] },
+    { "prefix": "shortcuts/slides/", "labelDomains": [], "e2eDomains": ["slides"] },
+    { "prefix": "shortcuts/note/", "labelDomains": [], "e2eDomains": ["note"] },
+    { "prefix": "shortcuts/event/", "labelDomains": [], "e2eDomains": ["event"] },
+
+    { "prefix": "skills/lark-im/", "labelDomains": ["im"], "e2eDomains": ["im"] },
+    { "prefix": "skills/lark-vc/", "labelDomains": ["vc"], "e2eDomains": ["vc"] },
+    { "prefix": "skills/lark-doc/", "labelDomains": ["ccm"], "e2eDomains": ["docs"] },
+    { "prefix": "skills/lark-wiki/", "labelDomains": ["ccm"], "e2eDomains": ["wiki"] },
+    { "prefix": "skills/lark-drive/", "labelDomains": ["ccm"], "e2eDomains": ["drive"] },
+    { "prefix": "skills/lark-sheets/", "labelDomains": ["ccm"], "e2eDomains": ["sheets"] },
+    { "prefix": "skills/lark-base/", "labelDomains": ["base"], "e2eDomains": ["base"] },
+    { "prefix": "skills/lark-mail/", "labelDomains": ["mail"], "e2eDomains": ["mail"] },
+    { "prefix": "skills/lark-calendar/", "labelDomains": ["calendar"], "e2eDomains": ["calendar"] },
+    { "prefix": "skills/lark-task/", "labelDomains": ["task"], "e2eDomains": ["task"] },
+    { "prefix": "skills/lark-contact/", "labelDomains": ["contact"], "e2eDomains": ["contact"] },
+    { "prefix": "skills/lark-apps/", "labelDomains": [], "e2eDomains": ["apps"] },
+    { "prefix": "skills/lark-markdown/", "labelDomains": [], "e2eDomains": ["markdown"] },
+    { "prefix": "skills/lark-minutes/", "labelDomains": [], "e2eDomains": ["minutes"] },
+    { "prefix": "skills/lark-okr/", "labelDomains": [], "e2eDomains": ["okr"] },
+    { "prefix": "skills/lark-slides/", "labelDomains": [], "e2eDomains": ["slides"] },
+    { "prefix": "skills/lark-note/", "labelDomains": [], "e2eDomains": ["note"] },
+    { "prefix": "skills/lark-event/", "labelDomains": [], "e2eDomains": ["event"] }
+  ],
+  "fullFallbackPrefixes": [
+    "shortcuts/common/",
+    "cmd/",
+    "internal/",
+    "pkg/",
+    "extension/",
+    "registry/",
+    "go.mod",
+    "go.sum",
+    "Makefile",
+    ".github/workflows/",
+    "scripts/"
+  ],
+  "skipPrefixes": [
+    "docs/",
+    ".changeset/"
+  ],
+  "skipSuffixes": [
+    ".md",
+    ".mdx",
+    ".txt",
+    ".rst"
+  ],
+  "skipFilenames": [
+    "readme.md",
+    "readme.zh.md",
+    "changelog.md",
+    "license",
+    "cla.md"
+  ]
+}
--- a/scripts/e2e_domains.js
+++ b/scripts/e2e_domains.js
@@ -0,0 +1,224 @@
+#!/usr/bin/env node
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+const fs = require("node:fs");
+const path = require("node:path");
+const { execFileSync } = require("node:child_process");
+const {
+  e2eDomainsForPath,
+  findPathMapping,
+  isSkippablePath,
+  matchesFullFallback,
+  normalizeRepoPath,
+} = require("./domain-map");
+
+const ROOT = process.env.E2E_DOMAINS_ROOT || path.join(__dirname, "..");
+process.chdir(ROOT);
+
+function execLines(command, args) {
+  return execFileSync(command, args, { encoding: "utf8" })
+    .split(/\r?\n/)
+    .map((line) => line.trim())
+    .filter(Boolean);
+}
+
+function modulePath() {
+  return execLines("go", ["list", "-m"])[0];
+}
+
+function rootPackage(moduleName) {
+  return `${moduleName}/tests/cli_e2e`;
+}
+
+function allLivePackages(moduleName) {
+  return execLines("go", ["list", "./tests/cli_e2e/..."])
+    .filter((pkg) => pkg !== rootPackage(moduleName))
+    .filter((pkg) => !pkg.endsWith("/demo"));
+}
+
+function allDryPackages(moduleName) {
+  return allLivePackages(moduleName);
+}
+
+const domainExistsCache = new Map();
+
+function domainExists(domain) {
+  if (domainExistsCache.has(domain)) {
+    return domainExistsCache.get(domain);
+  }
+  let exists = false;
+  try {
+    execFileSync("go", ["list", `./tests/cli_e2e/${domain}`], { stdio: "ignore" });
+    exists = true;
+  } catch {
+    exists = false;
+  }
+  domainExistsCache.set(domain, exists);
+  return exists;
+}
+
+function readChangedFiles() {
+  const changedFilesPath = process.env.E2E_DOMAIN_CHANGED_FILES;
+  if (changedFilesPath) {
+    return fs.readFileSync(changedFilesPath, "utf8")
+      .split(/\r?\n/)
+      .map(normalizeRepoPath)
+      .filter(Boolean);
+  }
+
+  if (process.env.GITHUB_EVENT_NAME !== "pull_request") {
+    return null;
+  }
+
+  const baseRef = process.env.GITHUB_BASE_REF || "main";
+  try {
+    execFileSync("git", ["rev-parse", "--verify", `origin/${baseRef}`], { stdio: "ignore" });
+    return execLines("git", ["diff", "--name-only", `origin/${baseRef}...HEAD`]).map(normalizeRepoPath);
+  } catch {
+    return null;
+  }
+}
+
+function addDomain(domains, domain) {
+  if (domain && domainExists(domain)) {
+    domains.add(domain);
+    return true;
+  }
+  return false;
+}
+
+function classifyPath(filePath, domains) {
+  const normalized = normalizeRepoPath(filePath);
+  if (!normalized) return { matched: false };
+
+  const e2eMatch = normalized.match(/^tests\/cli_e2e\/([^/]+)\//);
+  if (e2eMatch) {
+    const domain = e2eMatch[1];
+    if (domain === "demo") return { matched: false };
+    if (domainExists(domain)) {
+      addDomain(domains, domain);
+      return { matched: true };
+    }
+    if (isSkippablePath(normalized)) return { matched: false };
+    return { fullReason: `unknown CLI E2E domain path: ${normalized}` };
+  }
+
+  if (normalized.startsWith("tests/cli_e2e/")) {
+    return { fullReason: `shared CLI E2E harness changed: ${normalized}` };
+  }
+
+  if (matchesFullFallback(normalized)) {
+    return { fullReason: `shared/runtime path changed: ${normalized}` };
+  }
+
+  const mappedDomains = e2eDomainsForPath(normalized);
+  if (mappedDomains.length > 0) {
+    const missingDomains = [];
+    for (const domain of mappedDomains) {
+      if (!addDomain(domains, domain)) missingDomains.push(domain);
+    }
+    if (missingDomains.length > 0) {
+      return { fullReason: `mapped CLI E2E domain has no package: ${missingDomains.join(",")} (${normalized})` };
+    }
+    return { matched: true };
+  }
+
+  if (findPathMapping(normalized)) {
+    return { fullReason: `mapped path has no CLI E2E package: ${normalized}` };
+  }
+
+  if (normalized.match(/^shortcuts\/[^/]+\//) || normalized.match(/^skills\/lark-[^/]+\//)) {
+    return { fullReason: `unmapped CLI E2E domain path: ${normalized}` };
+  }
+
+  if (isSkippablePath(normalized)) return { matched: false };
+
+  return { fullReason: `unclassified path changed: ${normalized}` };
+}
+
+function resolveDomains(changedFiles) {
+  const moduleName = modulePath();
+  const rootDryPackage = rootPackage(moduleName);
+  if (changedFiles === null) {
+    return {
+      mode: "full",
+      reason: "non-pull_request run or unavailable diff",
+      domains: ["all"],
+      dryRootPackage: rootDryPackage,
+      dryPackages: allDryPackages(moduleName),
+      livePackages: allLivePackages(moduleName),
+    };
+  }
+
+  const domains = new Set();
+  let matchedRelevant = false;
+  let fullReason = "";
+
+  for (const file of changedFiles) {
+    const result = classifyPath(file, domains);
+    if (result.matched) matchedRelevant = true;
+    if (result.fullReason && !fullReason) fullReason = result.fullReason;
+  }
+
+  if (fullReason) {
+    return {
+      mode: "full",
+      reason: fullReason,
+      domains: ["all"],
+      dryRootPackage: rootDryPackage,
+      dryPackages: allDryPackages(moduleName),
+      livePackages: allLivePackages(moduleName),
+    };
+  }
+
+  if (matchedRelevant && domains.size > 0) {
+    const sortedDomains = [...domains].sort();
+    const packages = sortedDomains.map((domain) => `${moduleName}/tests/cli_e2e/${domain}`);
+    return {
+      mode: "subset",
+      reason: "business domain changes",
+      domains: sortedDomains,
+      dryRootPackage: rootDryPackage,
+      dryPackages: packages,
+      livePackages: packages,
+    };
+  }
+
+  return {
+    mode: "skip",
+    reason: "docs-only or no live CLI E2E impact",
+    domains: [],
+    dryRootPackage: "",
+    dryPackages: [],
+    livePackages: [],
+  };
+}
+
+function emit(resolved) {
+  const values = {
+    mode: resolved.mode,
+    reason: resolved.reason,
+    domains: resolved.domains.join(","),
+    dry_root_package: resolved.dryRootPackage,
+    dry_packages: resolved.dryPackages.join(" "),
+    live_packages: resolved.livePackages.join(" "),
+  };
+
+  const lines = Object.entries(values).map(([key, value]) => `${key}=${value}`);
+  console.log(lines.join("\n"));
+
+  if (process.env.GITHUB_OUTPUT) {
+    fs.appendFileSync(process.env.GITHUB_OUTPUT, `${lines.join("\n")}\n`);
+  }
+}
+
+if (require.main === module) {
+  emit(resolveDomains(readChangedFiles()));
+}
+
+module.exports = {
+  classifyPath,
+  readChangedFiles,
+  resolveDomains,
+};
--- a/scripts/e2e_domains.test.js
+++ b/scripts/e2e_domains.test.js
@@ -0,0 +1,94 @@
+// Copyright (c) 2026 Lark Technologies Pte. Ltd.
+// SPDX-License-Identifier: MIT
+
+const assert = require("node:assert/strict");
+const fs = require("node:fs");
+const os = require("node:os");
+const path = require("node:path");
+const { execFileSync } = require("node:child_process");
+const test = require("node:test");
+
+const scriptPath = path.join(__dirname, "e2e_domains.js");
+
+function parseOutput(raw) {
+  const result = {};
+  for (const line of raw.trim().split(/\r?\n/)) {
+    const idx = line.indexOf("=");
+    if (idx === -1) continue;
+    result[line.slice(0, idx)] = line.slice(idx + 1);
+  }
+  return result;
+}
+
+function runDomains(files) {
+  const dir = fs.mkdtempSync(path.join(os.tmpdir(), "e2e-domains-"));
+  const file = path.join(dir, "changed.txt");
+  fs.writeFileSync(file, `${files.join("\n")}\n`);
+  try {
+    return parseOutput(execFileSync(process.execPath, [scriptPath], {
+      cwd: path.join(__dirname, ".."),
+      encoding: "utf8",
+      env: { ...process.env, E2E_DOMAIN_CHANGED_FILES: file },
+    }));
+  } finally {
+    fs.rmSync(dir, { recursive: true, force: true });
+  }
+}
+
+test("maps shortcut changes to one business domain package", () => {
+  const output = runDomains(["shortcuts/im/messages/send.go"]);
+  assert.equal(output.mode, "subset");
+  assert.equal(output.domains, "im");
+  assert.match(output.dry_root_package, /github\.com\/larksuite\/cli\/tests\/cli_e2e$/);
+  assert.match(output.live_packages, /github\.com\/larksuite\/cli\/tests\/cli_e2e\/im/);
+  assert.doesNotMatch(output.live_packages, /github\.com\/larksuite\/cli\/tests\/cli_e2e\/drive/);
+});
+
+test("maps doc shortcuts to docs package", () => {
+  const output = runDomains(["shortcuts/doc/update.go"]);
+  assert.equal(output.mode, "subset");
+  assert.equal(output.domains, "docs");
+  assert.match(output.live_packages, /github\.com\/larksuite\/cli\/tests\/cli_e2e\/docs/);
+});
+
+test("maps direct e2e domain package changes", () => {
+  const output = runDomains(["tests/cli_e2e/drive/helpers.go"]);
+  assert.equal(output.mode, "subset");
+  assert.equal(output.domains, "drive");
+  assert.match(output.live_packages, /github\.com\/larksuite\/cli\/tests\/cli_e2e\/drive/);
+});
+
+test("falls back to full for shared e2e harness changes", () => {
+  const output = runDomains(["tests/cli_e2e/core.go"]);
+  assert.equal(output.mode, "full");
+  assert.equal(output.domains, "all");
+  assert.match(output.reason, /shared CLI E2E harness changed/);
+});
+
+test("falls back to full for runtime changes", () => {
+  const output = runDomains(["cmd/root.go"]);
+  assert.equal(output.mode, "full");
+  assert.equal(output.domains, "all");
+  assert.match(output.reason, /shared\/runtime path changed/);
+});
+
+test("skips docs-only changes", () => {
+  const output = runDomains(["docs/usage.md", "README.md"]);
+  assert.equal(output.mode, "skip");
+  assert.equal(output.domains, "");
+  assert.equal(output.dry_root_package, "");
+  assert.equal(output.live_packages, "");
+});
+
+test("uses shared map for skill domain changes", () => {
+  const output = runDomains(["skills/lark-sheets/SKILL.md"]);
+  assert.equal(output.mode, "subset");
+  assert.equal(output.domains, "sheets");
+  assert.match(output.live_packages, /github\.com\/larksuite\/cli\/tests\/cli_e2e\/sheets/);
+});
+
+test("falls back to full when a mapped path has no e2e package", () => {
+  const output = runDomains(["shortcuts/whiteboard/export.go"]);
+  assert.equal(output.mode, "full");
+  assert.match(output.reason, /unmapped CLI E2E domain path/);
+});
--- a/scripts/pr-labels/index.js
+++ b/scripts/pr-labels/index.js
@@ -4,6 +4,7 @@

 const fs = require("node:fs/promises");
 const path = require("node:path");
+const { labelDomainsForPath } = require("../domain-map");

 // ============================================================================
 // Constants & Configuration
@@ -35,33 +36,6 @@ const CORE_PREFIXES = ["internal/auth/", "internal/engine/", "internal/config/",
 const HEAD_BUSINESS_DOMAINS = new Set(["im", "contact", "ccm", "base", "docx"]);
 const LOW_RISK_TYPES = new Set(["docs", "ci", "test", "chore"]);

-// CODEOWNERS-based path to domain label mapping
-// Maps shortcuts and skills paths to business domain labels
-const PATH_TO_DOMAIN_MAP = {
-  // shortcuts
-  "shortcuts/im/": "im",
-  "shortcuts/vc/": "vc",
-  "shortcuts/calendar/": "calendar",
-  "shortcuts/doc/": "ccm",
-  "shortcuts/sheets/": "ccm",
-  "shortcuts/drive/": "ccm",
-  "shortcuts/wiki/": "ccm",
-  "shortcuts/base/": "base",
-  "shortcuts/mail/": "mail",
-  "shortcuts/task/": "task",
-  "shortcuts/contact/": "contact",
-  // skills
-  "skills/lark-im/": "im",
-  "skills/lark-vc/": "vc",
-  "skills/lark-doc/": "ccm",
-  "skills/lark-wiki/": "ccm",
-  "skills/lark-base/": "base",
-  "skills/lark-mail/": "mail",
-  "skills/lark-calendar/": "calendar",
-  "skills/lark-task/": "task",
-  "skills/lark-contact/": "contact",
-};
-
 const SENSITIVE_PATTERN = /(^|\/)(auth|permission|permissions|security)(\/|_|\.|$)/;

 const CLASS_STANDARDS = {
@@ -285,13 +259,7 @@ function skillDomainForPath(filePath) {

 // Get business domain label based on CODEOWNERS path mapping
 function getBusinessDomain(filePath) {
-  const normalized = normalizePath(filePath);
-  for (const [prefix, domain] of Object.entries(PATH_TO_DOMAIN_MAP)) {
-    if (normalized.startsWith(prefix)) {
-      return domain;
-    }
-  }
-  return "";
+  return labelDomainsForPath(filePath)[0] || "";
 }

 async function detectNewShortcutDomain(files) {
--- a/scripts/resolve-changed-from.test.sh
+++ b/scripts/resolve-changed-from.test.sh
@@ -8,7 +8,17 @@ repo_root="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
 script="$repo_root/scripts/resolve-changed-from.sh"

 tmp="${TMPDIR:-/tmp}/resolve-changed-from-test-$$"
-trap 'rm -rf "$tmp"' EXIT
+
+cleanup_tmp() {
+  local attempt
+  for attempt in 1 2 3; do
+    rm -rf "$tmp" && return 0
+    sleep 1
+  done
+  rm -rf "$tmp"
+}
+
+trap cleanup_tmp EXIT
 mkdir -p "$tmp"

 git_init() {
--- a/skills/lark-apps/references/lark-apps-openapi-key.md
+++ b/skills/lark-apps/references/lark-apps-openapi-key.md
@@ -76,4 +76,4 @@ CLI 提供三种互斥的 scope 表达方式：
 ## 不在本 skill 范围

 - OpenAPI spec 全量导出、实时日志 tail、Webhook 消费、多鉴权方式：本期不支持。
- 身份选择、权限不足处理（`permission_violations`→`console_url`）、exit-10 审批、通用"禁输出密钥"红线、高风险操作通用框架：见 [`../lark-shared/SKILL.md`](../lark-shared/SKILL.md)，不在此重复。
+- 身份选择、权限不足处理（`missing_scopes`→`console_url`）、exit-10 审批、通用"禁输出密钥"红线、高风险操作通用框架：见 [`../lark-shared/SKILL.md`](../lark-shared/SKILL.md)，不在此重复。
--- a/skills/lark-base/SKILL.md
+++ b/skills/lark-base/SKILL.md
@@ -85,7 +85,7 @@ metadata:
 ## 身份与权限降级

 - 默认显式使用 `--as user` 操作用户资源；只有用户明确要求应用身份时，才直接用 `--as bot`。
- user 身份报 scope/授权不足，或错误中包含 `permission_violations` / `hint`，先转 `lark-shared` 做用户授权恢复，不要直接降级 bot。
+- user 身份报 scope/授权不足，或错误中包含 `missing_scopes` / `hint`，先转 `lark-shared` 做用户授权恢复，不要直接降级 bot。
 - user 身份报资源级无访问且无授权恢复提示时，才可用 `--as bot` 重试一次；bot 仍失败就停止重试并按权限错误处理。
 - `91403` 或明确不可访问错误不要循环换身份重试。
 - `+base-create` / `+base-copy` 若用 bot 身份执行，关注返回中的 `permission_grant`，并把用户是否可打开新 Base 告知用户。
--- a/skills/lark-base/references/lark-base-dashboard-block-get-data.md
+++ b/skills/lark-base/references/lark-base-dashboard-block-get-data.md
@@ -98,16 +98,35 @@ lark-cli base +dashboard-block-get \

 ## 返回结构总览

-服务端响应外层仍然是标准 OpenAPI 包装：
+CLI 输出标准成功信封（判断成功用 `ok == true` 或退出码 0，不要找 `code == 0`）：

 ```json
 {
-  "code": 0,
-  "msg": "success",
+  "ok": true,
+  "identity": "user",
  "data": {
-    "dimensions": [...],
-    "measures": [...],
-    "main_data": [...]
+    "dimensions": [
+      {
+        "field_name": "地区",
+        "alias": "dim_region"
+      }
+    ],
+    "measures": [
+      {
+        "field_name": "销售额",
+        "alias": "me_sales"
+      }
+    ],
+    "main_data": [
+      {
+        "dim_region": {
+          "value": "华东"
+        },
+        "me_sales": {
+          "value": 12345
+        }
+      }
+    ]
  }
 }
 ```
--- a/skills/lark-base/references/lark-base-data-query.md
+++ b/skills/lark-base/references/lark-base-data-query.md
@@ -347,28 +347,51 @@ value 使用预定义关键字机制，第一个元素为字符串常量名称
 |------|------|------|------|
 | `format` | string | 是 | 固定为 `"flat"`，表示返回扁平化的对象数组 |

-## API 出参详情
+## CLI 出参详情

-**成功时：**
+**成功时**（stdout，判断成功用 `ok == true` 或退出码 0）：

 ```json
-{"code": 0, "data": {"main_data": [{"dim_city": {"value": "北京"}, "total_amount": {"value": 12345.00}}, ...]}, "msg": ""}
+{
+  "ok": true,
+  "identity": "user",
+  "data": {
+    "main_data": [
+      {
+        "dim_city": {
+          "value": "北京"
+        },
+        "total_amount": {
+          "value": 12345.00
+        }
+      },
+      {
+        "dim_city": {
+          "value": "上海"
+        },
+        "total_amount": {
+          "value": 6789.00
+        }
+      }
+    ]
+  }
+}
 ```

-**失败时：**
+**失败时**（stderr 类型化错误信封，非零退出码；`error.code` 是上游 API 错误码）：

 ```json
-{"code": 800004006, "data": {"error": {"code": 800004006, ...}}, "msg": "DSL validation failed"}
+{"ok": false, "identity": "user", "error": {"type": "api", "subtype": "...", "code": 800004006, "message": "DSL validation failed", "hint": "..."}}
 ```

 **Response 字段：**

 | 字段 | 类型 | 说明 |
 |------|------|------|
-| `code` | int | 状态码，0 为成功 |
-| `msg` | string | 错误信息 |
+| `ok` | bool | 是否成功 |
 | `data.main_data` | []object | 查询结果数组，每个元素为一行数据 |
-| `data.error` | object | 失败时的错误详情 |
+| `error.code` | int | 失败时的上游 API 错误码 |
+| `error.message` / `error.hint` | string | 失败原因与建议的恢复动作 |

 每行数据的字段值封装在 CellValue 中：

@@ -387,7 +410,7 @@ value 使用预定义关键字机制，第一个元素为字符串常量名称

 ## 返回值

-命令成功后输出 `data` 字段的内容：
+命令成功后，成功信封的 `data` 字段即查询结果：

 ```json
 {
--- a/skills/lark-drive/references/lark-drive-member-add.md
+++ b/skills/lark-drive/references/lark-drive-member-add.md
@@ -54,7 +54,7 @@ lark-cli drive +member-add \
 }
 ```

-批量部分失败时，`partial` 为 `true`，CLI 以非零退出码返回 `error.type=partial_failure`。检查 `error.detail` 中的 `requested_count`、`succeeded_count`、`members`、`missing_member_ids` 和可选的 `mismatched_member_ids`。响应顺序不影响匹配结果。
+批量部分失败时，`partial` 为 `true`，同一份结果以 `ok:false` 部分失败信封写到 **stdout**（stderr 不再输出单独的错误信封），CLI 以非零退出码结束。检查 `data` 中的 `requested_count`、`succeeded_count`、`members`、`missing_member_ids` 和可选的 `mismatched_member_ids`。响应顺序不影响匹配结果。

 ## 行为说明

--- a/skills/lark-drive/references/lark-drive-pull.md
+++ b/skills/lark-drive/references/lark-drive-pull.md
@@ -17,11 +17,11 @@
 | `summary.deleted_local` | 启用 `--delete-local --yes` 时删除的本地文件数 |
 | `items[]` | 每个文件的明细（`rel_path` / `file_token` / `source_id` / `action` / 失败时的 `error`） |

-`summary.failed > 0` 时命令以 **非零状态码**（`exit=1`，`error.type=partial_failure`）退出，且同一份 `summary + items` 会在 `error.detail` 里返回；脚本/agent 直接通过 exit code 判断成败即可，不需要再去解 `summary.failed`。
+`summary.failed > 0` 时命令以 **非零状态码**（`exit=1`）退出：同一份 `summary + items` 会以 `ok:false` 部分失败信封写到 **stdout**（字段在 `data.summary` / `data.items`，另附 `data.note` 说明失败情况），stderr 不再输出单独的错误信封；脚本/agent 直接通过 exit code 判断成败即可，不需要再去解 `summary.failed`。

 ## 远端同名文件冲突

-如果 Drive 中多个条目映射到同一个 `rel_path`，默认直接失败（`error.type=duplicate_remote_path`），且不会下载、覆盖或删除任何本地文件。只有“多个 `type=file` 同名”的场景支持显式策略；`file-folder` 这类异构冲突始终直接失败。
+如果 Drive 中多个条目映射到同一个 `rel_path`，默认直接失败（stderr 类型化错误信封：`error.type=validation`、`error.subtype=failed_precondition`，`error.params[]` 逐条列出冲突的 `rel_path` 及碰撞条目），且不会下载、覆盖或删除任何本地文件。只有“多个 `type=file` 同名”的场景支持显式策略；`file-folder` 这类异构冲突始终直接失败。

 | 策略 | 行为 |
 |------|------|
@@ -80,7 +80,7 @@ lark-cli drive +pull --local-dir ./repo --folder-token fldcnxxxxxxxxx \

 - `--delete-local`（无 `--yes`）→ Validate 直接报错：`--delete-local requires --yes`，没有任何下载、列表请求或删除发生。
 - `--delete-local --yes`，**且下载阶段全部成功** → 扫一遍 `--local-dir` 下所有常规文件，把不在云端清单里的逐个 `os.Remove`。**只删常规文件，不删目录**：远端文件夹被删除后，对应本地目录会保留空壳。
- `--delete-local --yes`，**但下载阶段有任何条目失败** → **跳过整个删除阶段**，命令以 `partial_failure` 非零退出。设计意图：避免出现"前面下载失败、后面继续删本地文件"的半同步状态；操作者修好下载错误后再重跑即可。
+- `--delete-local --yes`，**但下载阶段有任何条目失败** → **跳过整个删除阶段**，命令以 `ok:false` 部分失败结果非零退出。设计意图：避免出现"前面下载失败、后面继续删本地文件"的半同步状态；操作者修好下载错误后再重跑即可。
 - 远端同名文件冲突且使用默认 `fail` → 在下载阶段前失败，删除阶段不会运行。
 - 不传 `--delete-local` → `summary.deleted_local` 永远是 0；命令对本地"多余"文件视而不见。

--- a/skills/lark-drive/references/lark-drive-push.md
+++ b/skills/lark-drive/references/lark-drive-push.md
@@ -24,7 +24,7 @@

 ## 远端同名文件冲突

-如果 Drive 中多个条目映射到同一个 `rel_path`，默认直接失败（`error.type=duplicate_remote_path`），且不会上传、覆盖或进入 `--delete-remote` 删除阶段。只有“多个 `type=file` 同名”的场景支持显式策略；`file-folder` 这类异构冲突始终直接失败。
+如果 Drive 中多个条目映射到同一个 `rel_path`，默认直接失败（stderr 类型化错误信封：`error.type=validation`、`error.subtype=failed_precondition`，`error.params[]` 逐条列出冲突的 `rel_path` 及碰撞条目），且不会上传、覆盖或进入 `--delete-remote` 删除阶段。只有“多个 `type=file` 同名”的场景支持显式策略；`file-folder` 这类异构冲突始终直接失败。

 | 策略 | 行为 |
 |------|------|
--- a/skills/lark-drive/references/lark-drive-status.md
+++ b/skills/lark-drive/references/lark-drive-status.md
@@ -19,7 +19,7 @@

 ## 远端同名文件冲突

-如果 Drive 中多个条目映射到同一个 `rel_path`，`+status` 会在下载/hash 前直接失败，返回 `error.type=duplicate_remote_path`，并在 `error.detail.duplicates_remote[]` 中列出该路径下所有冲突条目的 `file_token`、`type`、名称、大小和时间字段；其中 `created_time`、`modified_time` 缺失时会省略，`size` 在缺失或为 `0` 时都可能被省略。不要把这种情况当成普通 `modified`；它表示同步域本身有歧义，需要先整理云端结构，或在 `+pull` / `+push` 中仅对“duplicate file”场景显式选择冲突策略。
+如果 Drive 中多个条目映射到同一个 `rel_path`，`+status` 会在下载/hash 前直接失败，在 stderr 返回类型化错误信封（`error.type=validation`、`error.subtype=failed_precondition`）；`error.params[]` 每条的 `name` 是冲突的 `rel_path`，`reason` 枚举该路径下所有碰撞条目（`type` + `file_token`）。不要把这种情况当成普通 `modified`；它表示同步域本身有歧义，需要先整理云端结构，或在 `+pull` / `+push` 中仅对“duplicate file”场景显式选择冲突策略（`error.hint` 也给出了同样的恢复选项）。

 ## 命令

@@ -76,20 +76,18 @@ lark-cli drive +status \
 ```json
 {
  "ok": false,
+  "identity": "user",
  "error": {
-    "type": "duplicate_remote_path",
-    "message": "multiple Drive entries map to the same rel_path",
-    "detail": {
-      "duplicates_remote": [
-        {
-          "rel_path": "dup.txt",
-          "entries": [
-            {"file_token": "<full_file_token>", "type": "file", "name": "dup.txt", "size": 5, "created_time": "1730000000", "modified_time": "1730000000"},
-            {"file_token": "<folder_token>", "type": "folder", "name": "dup.txt", "created_time": "1730000060", "modified_time": "1730000060"}
-          ]
-        }
-      ]
-    }
+    "type": "validation",
+    "subtype": "failed_precondition",
+    "message": "1 rel_path(s) map to multiple Drive entries",
+    "hint": "resolve the duplicate remote files first: re-run +pull with --on-duplicate-remote=rename (downloads each with a hashed suffix), or use --on-duplicate-remote=newest|oldest (supported by +pull/+sync/+push) to pick one, or delete the extra remote files; a plain retry will not help",
+    "params": [
+      {
+        "name": "dup.txt",
+        "reason": "2 Drive entries collide here: file <full_file_token>, folder <folder_token>"
+      }
+    ]
  }
 }
 ```
--- a/skills/lark-minutes/SKILL.md
+++ b/skills/lark-minutes/SKILL.md
@@ -120,9 +120,9 @@ lark-cli minutes +todo --minute-token <token> --as user --todos '[

 **更新 / 删除前**：先用 `minutes +detail --minute-tokens <token> --todo` 读取 `todos[].todo_id`（按 `content` 匹配目标条目；列表顺序不保证稳定，**不要**用"第 2 条"代替 `todo_id`）。

-**无编辑权限**：若 CLI 返回 `error.type=no_edit_permission`，表示对**这条妙记**没有编辑权，应请所有者授权；**不要**误走 `auth login --scope`。
+**无编辑权限**：若 CLI 返回稳定字段 `error.subtype=permission_denied`，且 `error.code` 为 `40005`（`+todo` / `+word-replace` 等编辑接口）或 `2091005`（如标题更新），表示对**这条妙记**没有编辑权，应请所有者授权；**不要**误走 `auth login --scope`。`error.message` 只作人读说明，不要用它做分支判断。

-**逐字稿关键词替换无命中**：`minutes +word-replace` 时，若 CLI 返回 `error.type=words_not_found`，表示传入的 `source_word` 在该妙记逐字稿中**一个都没匹配到**，未做任何替换。这是**参数问题不是权限问题**：先用 `minutes +detail --minute-tokens <token> --transcript` 读取当前逐字稿，核对 `source_word` 的精确写法与大小写后重试。
+**逐字稿关键词替换无命中**：`minutes +word-replace` 时，若 CLI 返回稳定字段 `error.code=40001` 且 `error.subtype=not_found`，表示传入的 `source_word` 在该妙记逐字稿中**一个都没匹配到**，未做任何替换。这是**参数问题不是权限问题**：先用 `minutes +detail --minute-tokens <token> --transcript` 读取当前逐字稿，核对 `source_word` 的精确写法与大小写后重试。`error.message` 只作人读说明，不要用它做分支判断。

 **替换 AI 总结全文**：见 [minutes +summary](references/lark-minutes-summary.md)。

--- a/skills/lark-minutes/references/lark-minutes-todo.md
+++ b/skills/lark-minutes/references/lark-minutes-todo.md
@@ -126,8 +126,8 @@ lark-cli minutes +todo --minute-token obcnxxxxxxxxxxxxxxxxxxxx --operation add -
 | 未指定操作 | 单条模式传 `--operation`，或批量传 `--todos` |
 | `--todos` 与单条 flags 冲突 | 二选一 |
 | `todos[i]` 校验失败 | 检查该条 `operation` 与字段组合 |
-| `error.type` = `no_edit_permission` | **妙记资源无编辑权**：向妙记所有者申请该妙记的编辑/协作权限；**不要**走 `auth login --scope` |
-| 缺少 OAuth scope（`permission_violations` 含 `minutes:minutes:update`） | `lark-cli auth login --scope "minutes:minutes:update"` |
+| `error.code=40005` 且 `error.subtype=permission_denied` | **妙记资源无编辑权**：向妙记所有者申请该妙记的编辑/协作权限；**不要**走 `auth login --scope`；`error.message` 只作人读说明 |
+| 缺少 OAuth scope（`error.missing_scopes` 含 `minutes:minutes:update`） | `lark-cli auth login --scope "minutes:minutes:update"` |

 ## 参考

--- a/skills/lark-shared/SKILL.md
+++ b/skills/lark-shared/SKILL.md
@@ -69,7 +69,7 @@ LARKSUITE_CLI_NO_UPDATE_NOTIFIER=1 LARKSUITE_CLI_NO_SKILLS_NOTIFIER=1 lark-cli a
 遇到权限相关错误时，**根据当前身份类型采取不同解决方案**。

 错误响应中包含关键信息：
- `permission_violations`：列出缺失的 scope (N选1)
+- `missing_scopes`：列出缺失的 scope (N选1)
 - `console_url`：飞书开发者后台的权限配置链接
 - `hint`：建议的修复命令

@@ -178,22 +178,22 @@ lark-cli 对高风险写操作（`risk: "high-risk-write"`）有强制确认门
 ```json
 {
  "ok": false,
+  "identity": "bot",
  "error": {
-    "type": "confirmation_required",
+    "type": "confirmation",
+    "subtype": "confirmation_required",
    "message": "drive +delete requires confirmation",
    "hint": "add --yes to confirm",
-    "risk": {
-      "level": "high-risk-write",
-      "action": "drive +delete"
-    }
+    "risk": "high-risk-write",
+    "action": "drive +delete"
  }
 }
 ```

 **遇到这种情况，不要当普通错误放弃。** 按以下流程处理：

-1. **识别**：看到子进程 exit code = `10` 且 stderr JSON 里 `error.type == "confirmation_required"`
-2. **向用户确认**：把 `error.risk.action` 和关键参数展示给用户，明确告知"这是高风险操作"，等待用户显式同意
+1. **识别**：看到子进程 exit code = `10` 且 stderr JSON 里 `error.type == "confirmation"`、`error.subtype == "confirmation_required"`
+2. **向用户确认**：把 `error.action`、`error.risk` 和关键参数展示给用户，明确告知"这是高风险操作"，等待用户显式同意
 3. **用户同意** → 在你**原始 argv 的末尾追加 `--yes`** 后重试
 4. **用户拒绝** → 终止流程，不要擅自改写参数或跳过门禁

--- a/skills/lark-slides/references/examples.md
+++ b/skills/lark-slides/references/examples.md
@@ -65,15 +65,15 @@ lark-cli slides xml_presentations get --as user --params '{

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "xml_presentation": {
      "presentation_id": "slides_example_presentation_id",
      "revision_id": 3,
      "content": "<presentation xmlns=\"http://www.larkoffice.com/sml/2.0\" height=\"540\" width=\"960\">...</presentation>"
    }
-  },
-  "msg": "success"
+  }
 }
 ```

@@ -94,12 +94,12 @@ lark-cli slides xml_presentation.slide create --as user --params '{

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "slide_id": "slide_example_id",
    "revision_id": 100
-  },
-  "msg": "success"
+  }
 }
 ```

@@ -116,11 +116,11 @@ lark-cli slides xml_presentation.slide delete --as user --params '{

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "revision_id": 101
-  },
-  "msg": "success"
+  }
 }
 ```

--- a/skills/lark-slides/references/lark-slides-screenshot.md
+++ b/skills/lark-slides/references/lark-slides-screenshot.md
@@ -66,7 +66,8 @@ lark-cli slides +screenshot --as user \

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "xml_presentation_id": "slides_example_presentation_id",
    "output_dir": ".lark-slides/screenshots",
@@ -79,8 +80,7 @@ lark-cli slides +screenshot --as user \
        "size": 12345
      }
    ]
-  },
-  "msg": "success"
+  }
 }
 ```

--- a/skills/lark-slides/references/lark-slides-xml-presentation-slide-create.md
+++ b/skills/lark-slides/references/lark-slides-xml-presentation-slide-create.md
@@ -141,12 +141,12 @@ lark-cli slides xml_presentation.slide create --as user \

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "slide_id": "slide_example_id",
    "revision_id": 100
-  },
-  "msg": "success"
+  }
 }
 ```

--- a/skills/lark-slides/references/lark-slides-xml-presentation-slide-delete.md
+++ b/skills/lark-slides/references/lark-slides-xml-presentation-slide-delete.md
@@ -61,11 +61,11 @@ lark-cli slides xml_presentation.slide delete --as user --params '{"xml_presenta

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "revision_id": 100
-  },
-  "msg": "success"
+  }
 }
 ```

--- a/skills/lark-slides/references/lark-slides-xml-presentation-slide-get.md
+++ b/skills/lark-slides/references/lark-slides-xml-presentation-slide-get.md
@@ -65,15 +65,15 @@ lark-cli slides xml_presentation.slide get --as user --params '{

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "slide": {
      "slide_id": "slide_example_id",
      "content": "<slide id=\"slide_example_id\"><style/><data>...</data></slide>"
    },
    "revision_id": 100
-  },
-  "msg": "success"
+  }
 }
 ```

--- a/skills/lark-slides/references/lark-slides-xml-presentation-slide-replace.md
+++ b/skills/lark-slides/references/lark-slides-xml-presentation-slide-replace.md
@@ -130,24 +130,28 @@ lark-cli slides xml_presentation.slide replace --as user --params '{

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "revision_id": 105
-  },
-  "msg": "success"
+  }
 }
 ```

 ### 失败（任一 part 失败，整批不生效）

-失败时返回非零错误码（如 3350001）。若后端能定位失败的 part，`data` 中可能附带：
+失败时命令以非零退出码结束，stderr 返回类型化错误信封（`error.type` / `error.subtype` / `error.code`（如 3350001）/ `error.message` / `error.hint`）。这个普通写命令的失败路径不会在 stdout 额外打印后端原始响应；脚本和 agent 应以退出码与 stderr 信封为准。

 ```json
 {
-  "code": 3350001,
-  "data": {
-    "failed_part_index": 0,
-    "failed_reason": "block not found"
+  "ok": false,
+  "identity": "user",
+  "error": {
+    "type": "api",
+    "subtype": "...",
+    "code": 3350001,
+    "message": "...",
+    "hint": "..."
  }
 }
 ```
@@ -155,8 +159,9 @@ lark-cli slides xml_presentation.slide replace --as user --params '{
 | 字段 | 类型 | 说明 |
 |------|------|------|
 | `data.revision_id` | integer | 成功时返回更新后最新版本号 |
-| `data.failed_part_index` | integer | 失败的 part 在 `parts` 数组中的索引（从 0 起） |
-| `data.failed_reason` | string | 失败原因 |
+| `error.code` | integer | 失败时的上游 API 错误码 |
+| `error.subtype` | string | 类型化错误子类，脚本可与 `error.code` 一起判断恢复动作 |
+| `error.hint` | string | 建议的恢复动作 |

 ## 常见错误

--- a/skills/lark-slides/references/lark-slides-xml-presentations-get.md
+++ b/skills/lark-slides/references/lark-slides-xml-presentations-get.md
@@ -56,15 +56,15 @@ lark-cli slides xml_presentations get --as user --params '{"xml_presentation_id"

 ```json
 {
-  "code": 0,
+  "ok": true,
+  "identity": "user",
  "data": {
    "xml_presentation": {
      "presentation_id": "slides_example_presentation_id",
      "revision_id": 1,
      "content": "<presentation xmlns=\"http://www.larkoffice.com/sml/2.0\" height=\"540\" width=\"960\">...</presentation>"
    }
-  },
-  "msg": "success"
+  }
 }
 ```

--- a/skills/lark-vc-agent/SKILL.md
+++ b/skills/lark-vc-agent/SKILL.md
@@ -18,7 +18,7 @@ metadata:
 ## 内测提示

 - 当前功能正在内测中，仅少数用户可用。忽略 [`../lark-shared/SKILL.md`](../lark-shared/SKILL.md) 中的权限申请流程，以下方的指引为准。
- 如果 `lark-cli` 提示 `missing required scope(s)` / `permission_violations`，不要走普通权限申请流程；先提示用户加入早鸟群确认内测权限已开通，再按“应用身份权限配置检查”处理应用权限、安装和数据范围。
+- 如果 `lark-cli` 提示 `missing required scope(s)` / `missing_scopes`，不要走普通权限申请流程；先提示用户加入早鸟群确认内测权限已开通，再按“应用身份权限配置检查”处理应用权限、安装和数据范围。
 - 如果 `lark-cli` 返回 `error.code=20017` / `ErrNotInGray`，提示用户加入早鸟群：`https://go.larkoffice.com/join-chat/2f4nb0e1-fe00-4f67-bed7-25beaf533fbd`。

 ## 定位
@@ -167,7 +167,7 @@ Shortcut 是对常用操作的高级封装（`lark-cli vc +<verb> [flags]`）。

 ## 应用身份权限配置检查

-应用身份 `--as bot` 报 `no permission`、`missing required scope(s)`、`permission_violations`、`ErrNotInGray` 或 `20017` 时，不要引导用户执行 `auth login`。按顺序检查：
+应用身份 `--as bot` 报 `no permission`、`missing required scope(s)`、`missing_scopes`、`ErrNotInGray` 或 `20017` 时，不要引导用户执行 `auth login`。按顺序检查：

 1. 以 CLI 返回的 metadata / error envelope 为准，确认提示的 VC Agent 相关权限已开通。常见读取 active meeting / events 需要会中事件读取权限；应用机器人入会 / 离会需要 bot 入会写权限。
 2. 应用已发布并安装到当前租户。
--- a/tests/cli_e2e/core.go
+++ b/tests/cli_e2e/core.go
@@ -24,7 +24,17 @@ import (
 const EnvBinaryPath = "LARK_CLI_BIN"
 const projectRootMarkerDir = "tests"
 const cliBinaryName = "lark-cli"
-const CleanupTimeout = 30 * time.Second
+
+const (
+	// CleanupTimeout is the outer teardown budget. Keep it above any
+	// per-resource wait so cleanup command retries still have room to run.
+	CleanupTimeout = 60 * time.Second
+
+	defaultRetryAttempts        = 4
+	defaultRetryInitialDelay    = time.Second
+	defaultRetryMaxDelay        = 6 * time.Second
+	defaultRetryBackoffMultiple = 2
+)

 func SkipWithoutUserToken(t *testing.T) {
 	t.Helper()
@@ -102,6 +112,34 @@ type Result struct {
 	RunErr     error
 }

+type cleanupWarningError struct {
+	err error
+}
+
+func (e *cleanupWarningError) Error() string {
+	return e.err.Error()
+}
+
+func (e *cleanupWarningError) Unwrap() error {
+	return e.err
+}
+
+// CleanupWarning marks a cleanup verification issue as non-fatal after the
+// destructive cleanup command itself has already succeeded.
+func CleanupWarning(err error) error {
+	if err == nil {
+		return nil
+	}
+	return &cleanupWarningError{err: err}
+}
+
+// IsCleanupWarning reports whether err should be logged without failing the
+// parent test.
+func IsCleanupWarning(err error) bool {
+	var warning *cleanupWarningError
+	return errors.As(err, &warning)
+}
+
 // RetryOptions configures retry behavior for flaky external API calls.
 type RetryOptions struct {
 	Attempts        int
@@ -111,8 +149,25 @@ type RetryOptions struct {
 	ShouldRetry     func(*Result) bool
 }

+// WaitOptions configures a bounded poll loop for eventually consistent cleanup
+// or verification checks.
+type WaitOptions struct {
+	Timeout      time.Duration
+	Interval     time.Duration
+	TimeoutError func() error
+}
+
 // RunCmd executes lark-cli and captures stdout/stderr/exit code.
+// Service errors that return {"error":{"retryable":true}} are retried with
+// bounded exponential backoff so individual tests do not need to remember
+// RunCmdWithRetry for normal transient server contention.
 func RunCmd(ctx context.Context, req Request) (*Result, error) {
+	return RunCmdWithRetry(ctx, req, RetryOptions{
+		ShouldRetry: ResultHasRetryableError,
+	})
+}
+
+func runCmdOnce(ctx context.Context, req Request) (*Result, error) {
 	binaryPath, err := ResolveBinaryPath(req)
 	if err != nil {
 		return nil, err
@@ -186,16 +241,16 @@ func buildCommandEnv(req Request) []string {
 // RunCmdWithRetry reruns a command when the result matches the configured retry condition.
 func RunCmdWithRetry(ctx context.Context, req Request, opts RetryOptions) (*Result, error) {
 	if opts.Attempts <= 0 {
-		opts.Attempts = 4
+		opts.Attempts = defaultRetryAttempts
 	}
 	if opts.InitialDelay <= 0 {
-		opts.InitialDelay = 1 * time.Second
+		opts.InitialDelay = defaultRetryInitialDelay
 	}
 	if opts.MaxDelay <= 0 {
-		opts.MaxDelay = 6 * time.Second
+		opts.MaxDelay = defaultRetryMaxDelay
 	}
 	if opts.BackoffMultiple <= 1 {
-		opts.BackoffMultiple = 2
+		opts.BackoffMultiple = defaultRetryBackoffMultiple
 	}
 	if opts.ShouldRetry == nil {
 		opts.ShouldRetry = func(result *Result) bool {
@@ -206,7 +261,7 @@ func RunCmdWithRetry(ctx context.Context, req Request, opts RetryOptions) (*Resu
 	delay := opts.InitialDelay
 	var lastResult *Result
 	for attempt := 1; attempt <= opts.Attempts; attempt++ {
-		result, err := RunCmd(ctx, req)
+		result, err := runCmdOnce(ctx, req)
 		if err != nil {
 			return nil, err
 		}
@@ -234,6 +289,63 @@ func RunCmdWithRetry(ctx context.Context, req Request, opts RetryOptions) (*Resu
 	return lastResult, nil
 }

+// ResultHasRetryableError reports whether lark-cli returned a structured
+// service error with error.retryable=true in either output stream.
+func ResultHasRetryableError(result *Result) bool {
+	if result == nil {
+		return false
+	}
+	return rawHasRetryableError(result.Stdout) || rawHasRetryableError(result.Stderr)
+}
+
+func rawHasRetryableError(raw string) bool {
+	payload := extractJSONPayload(raw)
+	if payload == "" {
+		return false
+	}
+	return gjson.Get(payload, "error.retryable").Bool()
+}
+
+// WaitForCondition polls condition until it returns true, an error, the context
+// is canceled, or the configured timeout expires.
+func WaitForCondition(ctx context.Context, opts WaitOptions, condition func() (bool, error)) error {
+	if condition == nil {
+		return errors.New("wait condition is nil")
+	}
+	if opts.Timeout <= 0 {
+		opts.Timeout = CleanupTimeout
+	}
+	if opts.Interval <= 0 {
+		opts.Interval = time.Second
+	}
+
+	deadline := time.NewTimer(opts.Timeout)
+	defer deadline.Stop()
+	ticker := time.NewTicker(opts.Interval)
+	defer ticker.Stop()
+
+	for {
+		done, err := condition()
+		if err != nil {
+			return err
+		}
+		if done {
+			return nil
+		}
+
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		case <-deadline.C:
+			if opts.TimeoutError != nil {
+				return opts.TimeoutError()
+			}
+			return fmt.Errorf("condition still false after %s", opts.Timeout)
+		case <-ticker.C:
+		}
+	}
+}
+
 // GenerateSuffix returns a high-entropy UTC timestamp suffix suitable for remote test resource names.
 func GenerateSuffix() string {
 	now := time.Now().UTC()
@@ -251,6 +363,10 @@ func ReportCleanupFailure(parentT *testing.T, prefix string, result *Result, err
 	parentT.Helper()

 	if err != nil {
+		if IsCleanupWarning(err) {
+			parentT.Logf("%s: %v", prefix, err)
+			return
+		}
 		parentT.Errorf("%s: %v", prefix, err)
 		return
 	}
@@ -271,26 +387,11 @@ func isCleanupSuppressedResult(result *Result) bool {
 		return false
 	}

-	raw := strings.TrimSpace(result.Stdout)
-	if raw == "" {
-		raw = strings.TrimSpace(result.Stderr)
+	payload := extractJSONPayload(result.Stdout)
+	if payload == "" {
+		payload = extractJSONPayload(result.Stderr)
 	}
-	if raw == "" {
-		return false
-	}
-
-	start := strings.LastIndex(raw, "\n{")
-	if start >= 0 {
-		start++
-	} else {
-		start = strings.Index(raw, "{")
-	}
-	if start < 0 {
-		return false
-	}
-
-	payload := raw[start:]
-	if !gjson.Valid(payload) {
+	if payload == "" {
 		return false
 	}

@@ -306,6 +407,32 @@ func isCleanupSuppressedResult(result *Result) bool {
 	return errType == "api_error" && (errCode == 800004135 || strings.Contains(errMessage, " limited"))
 }

+func extractJSONPayload(raw string) string {
+	raw = strings.TrimSpace(raw)
+	if raw == "" {
+		return ""
+	}
+	if gjson.Valid(raw) {
+		return raw
+	}
+
+	start := strings.LastIndex(raw, "\n{")
+	if start >= 0 {
+		start++
+	} else {
+		start = strings.Index(raw, "{")
+	}
+	if start < 0 {
+		return ""
+	}
+
+	payload := raw[start:]
+	if !gjson.Valid(payload) {
+		return ""
+	}
+	return payload
+}
+
 // ResolveBinaryPath finds the CLI binary path using request, env, then PATH.
 func ResolveBinaryPath(req Request) (string, error) {
 	if req.BinaryPath != "" {
--- a/tests/cli_e2e/core_test.go
+++ b/tests/cli_e2e/core_test.go
@@ -5,10 +5,12 @@ package clie2e

 import (
 	"context"
+	"errors"
 	"os"
 	"path/filepath"
 	"strings"
 	"testing"
+	"time"

 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
@@ -223,6 +225,88 @@ func TestRunCmd(t *testing.T) {
 		assert.NotContains(t, env, "LARKSUITE_CLI_APP_ID=cli_app_test")
 		assert.NotContains(t, env, "LARKSUITE_CLI_USER_ACCESS_TOKEN=uat_test")
 	})
+
+	t.Run("retries structured retryable service errors by default", func(t *testing.T) {
+		fake := newFakeCLI(t)
+		statePath := filepath.Join(t.TempDir(), "retry-count")
+		result, err := RunCmd(context.Background(), Request{
+			BinaryPath: fake.BinaryPath,
+			Args:       []string{"fail-once-retryable", statePath},
+		})
+		require.NoError(t, err)
+		result.AssertExitCode(t, 0)
+		result.AssertStdoutStatus(t, true)
+
+		countBytes, err := os.ReadFile(statePath)
+		require.NoError(t, err)
+		assert.Equal(t, "2\n", string(countBytes))
+	})
+
+	t.Run("does not retry non-retryable service errors by default", func(t *testing.T) {
+		fake := newFakeCLI(t)
+		statePath := filepath.Join(t.TempDir(), "retry-count")
+		result, err := RunCmd(context.Background(), Request{
+			BinaryPath: fake.BinaryPath,
+			Args:       []string{"always-non-retryable", statePath},
+		})
+		require.NoError(t, err)
+		result.AssertExitCode(t, 1)
+
+		countBytes, err := os.ReadFile(statePath)
+		require.NoError(t, err)
+		assert.Equal(t, "1\n", string(countBytes))
+	})
+}
+
+func TestRunCmdWithRetry(t *testing.T) {
+	t.Run("does not include RunCmd default retry as a nested retry", func(t *testing.T) {
+		fake := newFakeCLI(t)
+		statePath := filepath.Join(t.TempDir(), "retry-count")
+		result, err := RunCmdWithRetry(context.Background(), Request{
+			BinaryPath: fake.BinaryPath,
+			Args:       []string{"fail-once-retryable", statePath},
+		}, RetryOptions{
+			Attempts:     1,
+			InitialDelay: time.Millisecond,
+			MaxDelay:     time.Millisecond,
+			ShouldRetry:  ResultHasRetryableError,
+		})
+		require.NoError(t, err)
+		result.AssertExitCode(t, 1)
+
+		countBytes, err := os.ReadFile(statePath)
+		require.NoError(t, err)
+		assert.Equal(t, "1\n", string(countBytes))
+	})
+}
+
+func TestWaitForCondition(t *testing.T) {
+	t.Run("polls until condition succeeds", func(t *testing.T) {
+		attempts := 0
+		err := WaitForCondition(context.Background(), WaitOptions{
+			Timeout:  50 * time.Millisecond,
+			Interval: time.Millisecond,
+		}, func() (bool, error) {
+			attempts++
+			return attempts == 2, nil
+		})
+
+		require.NoError(t, err)
+		assert.Equal(t, 2, attempts)
+	})
+
+	t.Run("returns custom timeout error", func(t *testing.T) {
+		wantErr := errors.New("still visible")
+		err := WaitForCondition(context.Background(), WaitOptions{
+			Timeout:      time.Millisecond,
+			Interval:     time.Millisecond,
+			TimeoutError: func() error { return wantErr },
+		}, func() (bool, error) {
+			return false, nil
+		})
+
+		assert.ErrorIs(t, err, wantErr)
+	})
 }

 type fakeCLI struct {
@@ -260,6 +344,35 @@ if [ "$1" = "emit-stdin" ]; then
  exit 0
 fi

+if [ "$1" = "fail-once-retryable" ]; then
+  state="$2"
+  count=0
+  if [ -f "$state" ]; then
+    count="$(cat "$state")"
+  fi
+  count=$((count + 1))
+  echo "$count" > "$state"
+  if [ "$count" -eq 1 ]; then
+    echo "Deleting folder fake..." >&2
+    echo '{"ok":false,"error":{"type":"api","code":1061045,"message":"resource contention occurred, please retry.","retryable":true}}' >&2
+    exit 1
+  fi
+  echo '{"ok":true}'
+  exit 0
+fi
+
+if [ "$1" = "always-non-retryable" ]; then
+  state="$2"
+  count=0
+  if [ -f "$state" ]; then
+    count="$(cat "$state")"
+  fi
+  count=$((count + 1))
+  echo "$count" > "$state"
+  echo '{"ok":false,"error":{"type":"api","code":123,"message":"validation failed","retryable":false}}' >&2
+  exit 1
+fi
+
 exit_code=0
 while [ "$#" -gt 0 ]; do
  case "$1" in
--- a/tests/cli_e2e/docs/coverage.md
+++ b/tests/cli_e2e/docs/coverage.md
@@ -21,7 +21,7 @@
 | Status | Cmd | Type | Testcase | Key parameter shapes | Notes / uncovered reason |
 | --- | --- | --- | --- | --- | --- |
 | ✓ | 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 +fetch | shortcut | docs_fetch_dryrun_test.go::TestDocsFetchDryRunIgnoresAPIVersionCompatFlag; 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`; `--api-version v1` compatibility flag still dry-runs the v2 fetch endpoint | |
 | ✓ | 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` |
--- a/tests/cli_e2e/docs/docs_fetch_dryrun_test.go
+++ b/tests/cli_e2e/docs/docs_fetch_dryrun_test.go
@@ -1,7 +1,7 @@
 // Copyright (c) 2026 Lark Technologies Pte. Ltd.
 // SPDX-License-Identifier: MIT

-package doc
+package docs

 import (
 	"context"
@@ -23,7 +23,7 @@ func TestDocsFetchDryRunIgnoresAPIVersionCompatFlag(t *testing.T) {
 		Args: []string{
 			"docs", "+fetch",
 			"--doc", "doxcnDryRunCompat",
-			"--api-version", "legacy",
+			"--api-version", "v1",
 			"--dry-run",
 		},
 		DefaultAs: "bot",
--- a/tests/cli_e2e/drive/coverage.md
+++ b/tests/cli_e2e/drive/coverage.md
@@ -9,7 +9,7 @@
 - TestDrive_FilesCreateFolderWorkflow: proves `drive files create_folder` in `create_folder as bot`; helper asserts the returned folder token and registers best-effort cleanup via `drive files delete`.
 - TestDrive_StatusWorkflow: proves `drive +status` against a real Drive folder. Seeds the remote side via `drive +upload` (`unchanged.txt`, `modified.txt`, `remote-only.txt`), seeds local files with the matching/diverging contents, and asserts every output bucket (`unchanged`, `modified`, `new_local`, `new_remote`) holds exactly the expected `rel_path` and `file_token`. Cleans up uploaded files and the parent folder via best-effort cleanup hooks.
 - TestDrive_UploadWorkflow: proves `drive +upload` against the real backend in both create and overwrite modes. First uploads a fresh file into a temporary Drive folder, then re-uploads new bytes with `--file-token` against the returned token, asserts the overwrite keeps the token stable, and finally downloads the file to confirm the remote content changed.
- TestDrive_DuplicateRemoteWorkflow: proves the duplicate-remote workflows against the real backend. One subtest uploads two same-name files into the same Drive folder and asserts `drive +status` and default `drive +pull` both fail with `duplicate_remote_path`, while `drive +pull --on-duplicate-remote=rename` succeeds, downloads both files, and writes a hashed renamed sibling locally. The other subtest uploads duplicate remote files, runs `drive +push --on-duplicate-remote=newest --if-exists=overwrite --delete-remote --yes`, and then re-runs `drive +status` to prove the mirror converged to a single unchanged `dup.txt`.
+- TestDrive_DuplicateRemoteWorkflow: proves the duplicate-remote workflows against the real backend. One subtest uploads two same-name files into the same Drive folder and asserts `drive +status` and default `drive +pull` both fail with a typed validation error for the duplicate rel_path, while `drive +pull --on-duplicate-remote=rename` succeeds, downloads both files, and writes a hashed renamed sibling locally. The other subtest uploads duplicate remote files, runs `drive +push --on-duplicate-remote=newest --if-exists=overwrite --delete-remote --yes`, and then re-runs `drive +status` to prove the mirror converged to a single unchanged `dup.txt`.
 - TestDrive_ApplyPermissionDryRun / TestDrive_ApplyPermissionDryRunRejectsFullAccess: dry-run coverage for `drive +apply-permission`; asserts URL→type inference for docx/sheet/slides, explicit `--type` overriding URL inference when both a recognized URL and `--type` are supplied, bare-token + explicit `--type` path, request method/URL/type-query/perm/remark body shape, optional `remark` omission when unset, and client-side rejection of `--perm full_access`. Runs without hitting the live API.
 - TestDriveAddCommentDryRun_File / TestDriveAddCommentDryRun_Base: dry-run coverage for `drive +add-comment` on supported Drive file and Base targets; pins the `metas.batch_query -> files/:token/new_comments` file chain, Base `file_type=bitable`, and Base anchor fields.
 - TestDriveAddCommentMarkdownFileWorkflow: opt-in live workflow skeleton for the same path, gated by `LARK_DRIVE_MD_COMMENT_E2E=1`.
--- a/tests/cli_e2e/drive/helpers.go
+++ b/tests/cli_e2e/drive/helpers.go
@@ -14,6 +14,18 @@ import (
 	"github.com/tidwall/gjson"
 )

+const (
+	driveDeleteVisibilityTimeout = 30 * time.Second
+	driveDeleteVisibilityPoll    = 3 * time.Second
+)
+
+var driveDeleteVisibilityWait = clie2e.WaitOptions{
+	// This wait only covers the post-delete visibility lag after Drive accepts
+	// deletion. The delete command itself is bounded by clie2e.CleanupContext.
+	Timeout:  driveDeleteVisibilityTimeout,
+	Interval: driveDeleteVisibilityPoll,
+}
+
 // CreateDriveFolder creates a Drive folder, optionally under a parent folder, and
 // deletes it during parent cleanup.
 func CreateDriveFolder(t *testing.T, parentT *testing.T, ctx context.Context, name string, defaultAs string, parentFolderToken string) string {
@@ -60,14 +72,18 @@ func CreateDriveFolder(t *testing.T, parentT *testing.T, ctx context.Context, na
 // returned a suppressed not_found or partial API error but the resource still
 // exists.
 func DeleteDriveResourceAndVerify(ctx context.Context, token, docType, defaultAs string) (*clie2e.Result, error) {
+	return deleteDriveResourceAndVerify(ctx, token, docType, defaultAs, driveDeleteVisibilityWait)
+}
+
+func deleteDriveResourceAndVerify(ctx context.Context, token, docType, defaultAs string, visibilityWait clie2e.WaitOptions) (*clie2e.Result, error) {
 	if defaultAs == "" {
 		defaultAs = "bot"
 	}

-	deleteResult, deleteErr := clie2e.RunCmdWithRetry(ctx, clie2e.Request{
+	deleteResult, deleteErr := clie2e.RunCmd(ctx, clie2e.Request{
 		Args:      []string{"drive", "+delete", "--file-token", token, "--type", docType, "--yes"},
 		DefaultAs: defaultAs,
-	}, clie2e.RetryOptions{})
+	})
 	if deleteErr != nil || deleteResult == nil {
 		return deleteResult, deleteErr
 	}
@@ -82,35 +98,21 @@ func DeleteDriveResourceAndVerify(ctx context.Context, token, docType, defaultAs
 		}
 		return deleteResult, fmt.Errorf("drive resource %s/%s still exists after delete failed: exit=%d stdout=%s stderr=%s", docType, token, deleteResult.ExitCode, deleteResult.Stdout, deleteResult.Stderr)
 	}
-	if err := WaitDriveResourceDeleted(ctx, token, docType, defaultAs); err != nil {
-		return deleteResult, err
+	if err := waitDriveResourceDeleted(ctx, token, docType, defaultAs, visibilityWait); err != nil {
+		return deleteResult, clie2e.CleanupWarning(
+			fmt.Errorf("drive resource %s/%s still visible after accepted delete: %w", docType, token, err),
+		)
 	}
 	return deleteResult, nil
 }

-func WaitDriveResourceDeleted(ctx context.Context, token, docType, defaultAs string) error {
-	deadline := time.NewTimer(20 * time.Second)
-	defer deadline.Stop()
-	ticker := time.NewTicker(time.Second)
-	defer ticker.Stop()
-
-	for {
-		deleted, err := IsDriveResourceDeleted(ctx, token, docType, defaultAs)
-		if err != nil {
-			return err
-		}
-		if deleted {
-			return nil
-		}
-
-		select {
-		case <-ctx.Done():
-			return ctx.Err()
-		case <-deadline.C:
-			return fmt.Errorf("drive resource %s/%s still exists after delete", docType, token)
-		case <-ticker.C:
-		}
+func waitDriveResourceDeleted(ctx context.Context, token, docType, defaultAs string, opts clie2e.WaitOptions) error {
+	opts.TimeoutError = func() error {
+		return fmt.Errorf("drive resource %s/%s still exists %s after delete", docType, token, opts.Timeout)
 	}
+	return clie2e.WaitForCondition(ctx, opts, func() (bool, error) {
+		return IsDriveResourceDeleted(ctx, token, docType, defaultAs)
+	})
 }

 func IsDriveResourceDeleted(ctx context.Context, token, docType, defaultAs string) (bool, error) {
--- a/tests/cli_e2e/drive/helpers_test.go
+++ b/tests/cli_e2e/drive/helpers_test.go
@@ -5,8 +5,13 @@ package drive

 import (
 	"context"
+	"os"
+	"path/filepath"
 	"testing"
+	"time"

+	clie2e "github.com/larksuite/cli/tests/cli_e2e"
+	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )

@@ -16,3 +21,59 @@ func createDriveFolder(t *testing.T, parentT *testing.T, ctx context.Context, na
 	require.NotEmpty(t, folderToken)
 	return folderToken
 }
+
+func TestDeleteDriveResourceAndVerify(t *testing.T) {
+	t.Run("successful delete with stale meta returns cleanup warning", func(t *testing.T) {
+		fake := mustWriteDriveCleanupFakeCLI(t)
+		t.Setenv(clie2e.EnvBinaryPath, fake)
+
+		result, err := deleteDriveResourceAndVerify(context.Background(), "fld_stale", "folder", "bot", clie2e.WaitOptions{
+			Timeout:  10 * time.Millisecond,
+			Interval: time.Millisecond,
+		})
+		require.NotNil(t, result)
+		assert.Equal(t, 0, result.ExitCode)
+		require.Error(t, err)
+		assert.True(t, clie2e.IsCleanupWarning(err), "err: %v", err)
+	})
+
+	t.Run("failed delete with existing meta remains fatal", func(t *testing.T) {
+		fake := mustWriteDriveCleanupFakeCLI(t)
+		t.Setenv(clie2e.EnvBinaryPath, fake)
+		t.Setenv("FAKE_DRIVE_DELETE_EXIT", "1")
+
+		result, err := DeleteDriveResourceAndVerify(context.Background(), "fld_existing", "folder", "bot")
+		require.NotNil(t, result)
+		assert.Equal(t, 1, result.ExitCode)
+		require.Error(t, err)
+		assert.False(t, clie2e.IsCleanupWarning(err), "err: %v", err)
+		assert.Contains(t, err.Error(), "still exists after delete failed")
+	})
+}
+
+func mustWriteDriveCleanupFakeCLI(t *testing.T) string {
+	t.Helper()
+
+	script := `#!/bin/sh
+if [ "$1" = "drive" ] && [ "$2" = "+delete" ]; then
+  if [ "${FAKE_DRIVE_DELETE_EXIT:-0}" != "0" ]; then
+    echo '{"ok":false,"error":{"type":"api","message":"delete failed"}}' >&2
+    exit "$FAKE_DRIVE_DELETE_EXIT"
+  fi
+  echo '{"ok":true}'
+  exit 0
+fi
+
+if [ "$1" = "api" ] && [ "$2" = "post" ] && [ "$3" = "/open-apis/drive/v1/metas/batch_query" ]; then
+  echo '{"ok":true,"data":{"metas":[{"url":"https://example.com/still-visible"}]}}'
+  exit 0
+fi
+
+echo "unexpected fake CLI args: $*" >&2
+exit 2
+`
+
+	binaryPath := filepath.Join(t.TempDir(), "fake-lark-cli")
+	require.NoError(t, os.WriteFile(binaryPath, []byte(script), 0o755))
+	return binaryPath
+}
--- a/tests/cli_e2e/wiki/helpers_test.go
+++ b/tests/cli_e2e/wiki/helpers_test.go
@@ -5,6 +5,7 @@ package wiki

 import (
 	"context"
+	"errors"
 	"fmt"
 	"strings"
 	"testing"
@@ -203,6 +204,11 @@ type wikiNodeInfo struct {
 	ObjType   string
 }

+const (
+	wikiDeleteVisibilityTimeout = 30 * time.Second
+	wikiDeleteVisibilityPoll    = 3 * time.Second
+)
+
 // deleteWikiNodeAndVerify removes a wiki node, then polls get_node until the
 // original node token is gone. Wiki cleanup cannot use drive +delete because
 // wiki origin nodes need the backing obj_token and parent nodes must delete
@@ -333,28 +339,34 @@ func listWikiNodeChildren(ctx context.Context, spaceID, parentNodeToken string)
 }

 func waitWikiNodeDeleted(ctx context.Context, nodeToken string) error {
-	deadline := time.NewTimer(20 * time.Second)
-	defer deadline.Stop()
-	ticker := time.NewTicker(time.Second)
-	defer ticker.Stop()
+	var lastTransientErr error

-	for {
+	opts := clie2e.WaitOptions{
+		Timeout:  wikiDeleteVisibilityTimeout,
+		Interval: wikiDeleteVisibilityPoll,
+		TimeoutError: func() error {
+			if lastTransientErr != nil {
+				return fmt.Errorf("wiki node %s delete verification kept hitting transient errors: %w", nodeToken, lastTransientErr)
+			}
+			return fmt.Errorf("wiki node %s still exists after delete", nodeToken)
+		},
+	}
+
+	return clie2e.WaitForCondition(ctx, opts, func() (bool, error) {
 		deleted, err := isWikiNodeDeleted(ctx, nodeToken)
 		if err != nil {
-			return err
+			if isWikiVerifyTransientError(err) {
+				lastTransientErr = err
+				return false, nil
+			} else {
+				return false, err
+			}
 		}
 		if deleted {
-			return nil
+			return true, nil
 		}
-
-		select {
-		case <-ctx.Done():
-			return ctx.Err()
-		case <-deadline.C:
-			return fmt.Errorf("wiki node %s still exists after delete", nodeToken)
-		case <-ticker.C:
-		}
-	}
+		return false, nil
+	})
 }

 func isWikiNodeDeleted(ctx context.Context, nodeToken string) (bool, error) {
@@ -375,9 +387,31 @@ func isWikiNodeDeleted(ctx context.Context, nodeToken string) (bool, error) {
 	if isWikiNodeDeletedResult(result) {
 		return true, nil
 	}
+	if isWikiVerifyTransientResult(result) {
+		return false, wikiVerifyTransientError{
+			err: fmt.Errorf("verify wiki node %s after delete hit transient response: exit=%d stdout=%s stderr=%s", nodeToken, result.ExitCode, result.Stdout, result.Stderr),
+		}
+	}
 	return false, fmt.Errorf("verify wiki node %s after delete: exit=%d stdout=%s stderr=%s", nodeToken, result.ExitCode, result.Stdout, result.Stderr)
 }

+type wikiVerifyTransientError struct {
+	err error
+}
+
+func (e wikiVerifyTransientError) Error() string {
+	return e.err.Error()
+}
+
+func (e wikiVerifyTransientError) Unwrap() error {
+	return e.err
+}
+
+func isWikiVerifyTransientError(err error) bool {
+	var transient wikiVerifyTransientError
+	return err != nil && errors.As(err, &transient)
+}
+
 func wikiAPISuccess(stdout string) bool {
 	if ok := gjson.Get(stdout, "ok"); ok.Exists() {
 		return ok.Bool()
@@ -404,6 +438,55 @@ func isWikiNodeDeletedResult(result *clie2e.Result) bool {
 		strings.Contains(combined, "not found")
 }

+func isWikiVerifyTransientResult(result *clie2e.Result) bool {
+	if result == nil {
+		return false
+	}
+	payload := result.Stdout
+	if strings.TrimSpace(payload) == "" {
+		payload = result.Stderr
+	}
+	if gjson.Get(payload, "error.type").String() != "internal" ||
+		gjson.Get(payload, "error.subtype").String() != "invalid_response" {
+		return false
+	}
+	message := strings.ToLower(gjson.Get(payload, "error.message").String())
+	return strings.Contains(message, "http 429") ||
+		strings.Contains(message, "http 500") ||
+		strings.Contains(message, "http 502") ||
+		strings.Contains(message, "http 503") ||
+		strings.Contains(message, "http 504")
+}
+
+func TestWikiVerifyTransientResult(t *testing.T) {
+	t.Run("matches invalid response from transient http status", func(t *testing.T) {
+		result := &clie2e.Result{
+			ExitCode: 5,
+			Stderr:   `{"ok":false,"error":{"type":"internal","subtype":"invalid_response","message":"SDK returned an invalid JSON response: failed to parse TAT response (HTTP 429): invalid character 'r' looking for beginning of value"}}`,
+		}
+
+		require.True(t, isWikiVerifyTransientResult(result))
+	})
+
+	t.Run("does not match unrelated invalid response", func(t *testing.T) {
+		result := &clie2e.Result{
+			ExitCode: 5,
+			Stderr:   `{"ok":false,"error":{"type":"internal","subtype":"invalid_response","message":"SDK returned an invalid JSON response: malformed body"}}`,
+		}
+
+		require.False(t, isWikiVerifyTransientResult(result))
+	})
+
+	t.Run("does not match api errors", func(t *testing.T) {
+		result := &clie2e.Result{
+			ExitCode: 1,
+			Stderr:   `{"ok":false,"error":{"type":"api","subtype":"conflict","message":"resource contention occurred, please retry","retryable":true}}`,
+		}
+
+		require.False(t, isWikiVerifyTransientResult(result))
+	})
+}
+
 func findWikiNodeByToken(t *testing.T, ctx context.Context, spaceID string, nodeToken string, parentNodeTokens ...string) gjson.Result {
 	t.Helper()