* docs: warn about unofficial PyPI packages and recommend version verification (#1982)
Clarify that only packages from github/spec-kit are official, and add
`specify version` as a post-install verification step to help users
catch accidental installation of an unrelated package with the same name.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): auto-correct legacy command names instead of hard-failing (#2017)
Community extensions that predate the strict naming requirement use two
common legacy formats ('speckit.command' and 'extension.command').
Instead of rejecting them outright, auto-correct to the required
'speckit.{extension}.{command}' pattern and emit a compatibility warning
so authors know they need to update their manifest. Names that cannot be
safely corrected (e.g. single-segment names) still raise ValidationError.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(tests): isolate preset catalog search test from community catalog network calls
test_search_with_cached_data asserted exactly 2 results but was getting 4
because _get_merged_packs() queries the full built-in catalog stack
(default + community). The community catalog had no local cache and hit
the network, returning real presets. Writing a project-level
preset-catalogs.yml that pins the test to the default URL only makes
the count assertions deterministic.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): extend auto-correction to aliases (#2017)
The upstream #1994 added alias validation in _collect_manifest_command_names,
which also rejected legacy 2-part alias names (e.g. 'speckit.verify').
Extend the same auto-correction logic from _validate() to cover aliases,
so both 'speckit.command' and 'extension.command' alias formats are
corrected to 'speckit.{ext_id}.{command}' with a compatibility warning
instead of hard-failing.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): address PR review feedback (#2017)
- _try_correct_command_name: only correct 'X.Y' to 'speckit.ext_id.Y'
when X matches ext_id, preventing misleading warnings followed by
install failure due to namespace mismatch
- _validate: add aliases type/string guards matching _collect_manifest
_command_names defensive checks
- _validate: track command renames and rewrite any hook.*.command
references that pointed at a renamed command, emitting a warning
- test: fix test_command_name_autocorrect_no_speckit_prefix to use
ext_id matching the legacy namespace; add namespace-mismatch test
- test: replace redundant preset-catalogs.yml isolation with
monkeypatch.delenv("SPECKIT_PRESET_CATALOG_URL") so the env var
cannot bypass catalog restriction in CI environments
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* Update docs/installation.md
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* fix(extensions): warn when hook command refs are silently canonicalized; fix grammar
- Hook rewrites (alias-form or rename-map) now always emit a warning so
extension authors know to update their manifests. Previously only
rename-map rewrites produced a warning; pure alias-form lifts were
silent.
- Pluralize "command/commands" in the uninstall confirmation message so
single-command extensions no longer print "1 commands".
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): raise ValidationError for non-dict hook entries
Silently skipping non-dict hook entries left them in manifest.hooks,
causing HookExecutor.register_hooks() to crash with AttributeError
when it called hook_config.get() on a non-mapping value.
Also updates PR description to accurately reflect the implementation
(no separate _try_correct_alias_name helper; aliases use the same
_try_correct_command_name path).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): derive remove cmd_count from registry, fix wording
Previously cmd_count used len(ext_manifest.commands) which only counted
primary commands and missed aliases. The registry's registered_commands
already tracks every command name (primaries + aliases) per agent, so
max(len(v) for v in registered_commands.values()) gives the correct
total.
Also changes "from AI agent" → "across AI agents" since remove()
unregisters commands from all detected agents.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): distinguish missing vs empty registered_commands in remove prompt
Using get() without a default lets us tell apart:
- key missing (legacy registry entry) → fall back to manifest count
- key present but empty dict (installed with no agent dirs) → show 0
Previously the truthiness check `if registered_commands and ...` treated
both cases the same, so an empty dict fell back to len(manifest.commands)
and overcounted commands that would actually be removed.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): clarify removal prompt wording to 'per agent'
'across AI agents' implied a total count, but cmd_count uses max()
across agents (per-agent count). Using sum() would double-count since
users think in logical commands, not per-agent files. 'per agent'
accurately describes what the number represents.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): clarify cmd_count comment — per-agent max, not total
The comment said 'covers all agents' implying a total, but cmd_count uses
max() across agents (per-agent count). Updated comment to explain the
max() choice and why sum() would double-count.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* test(extensions): add CLI tests for remove confirmation pluralization
Adds TestExtensionRemoveCLI with two CliRunner tests:
- singular: 1 registered command → '1 command per agent'
- plural: 2 registered commands → '2 commands per agent'
These prevent regressions on the cmd_count pluralization logic
and the 'per agent' wording introduced in this PR.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(agents): remove orphaned SKILL.md parent dirs on unregister
For SKILL.md-based agents (codex, kimi), each command lives in its own
subdirectory (e.g. .agents/skills/speckit-ext-cmd/SKILL.md). The previous
unregister_commands() only unlinked the file, leaving an empty parent dir.
Now attempts rmdir() on the parent when it differs from the agent commands
dir. OSError is silenced so non-empty dirs (e.g. user files) are safely left.
Adds test_unregister_skill_removes_parent_directory to cover this.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): drop alias pattern enforcement from _validate()
Aliases are intentionally free-form to preserve community extension
compatibility (e.g. 'speckit.verify' short aliases used by spec-kit-verify
and other existing extensions). This aligns _validate() with the intent of
upstream commit 4deb90f (fix: restore alias compatibility, #2110/#2125).
Only type and None-normalization checks remain for aliases. Pattern
enforcement continues for primary command names only.
Updated tests to verify free-form aliases pass through unchanged with
no warnings instead of being auto-corrected.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(extensions): guard against non-dict command entries in _validate()
If provides.commands contains a non-mapping entry (e.g. an int or string),
'name' not in cmd raises TypeError instead of a user-facing ValidationError.
Added isinstance(cmd, dict) check at the top of the loop.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: iamaeroplane <michal.bachorik@gmail.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* feat(cli): embed core pack in wheel + offline-first init (#1711, #1752)
Bundle templates, commands, and scripts inside the specify-cli wheel so
that `specify init` works without any network access by default.
Changes:
- pyproject.toml: add hatchling force-include for core_pack assets; bump
version to 0.2.1
- __init__.py: add _locate_core_pack(), _generate_agent_commands() (Python
port of generate_commands() shell function), and scaffold_from_core_pack();
modify init() to scaffold from bundled assets by default; add --from-github
flag to opt back in to the GitHub download path
- release.yml: build wheel during CI release job
- create-github-release.sh: attach .whl as a release asset
- docs/installation.md: add Enterprise/Air-Gapped Installation section
- README.md: add Option 3 enterprise install with accurate offline story
Closes#1711
Addresses #1752
* fix(tests): update kiro alias test for offline-first scaffold path
* feat(cli): invoke bundled release script at runtime for offline scaffold
- Embed release scripts (bash + PowerShell) in wheel via pyproject.toml
- Replace Python _generate_agent_commands() with subprocess invocation of
the canonical create-release-packages.sh, guaranteeing byte-for-byte
parity between 'specify init --offline' and GitHub release ZIPs
- Fix macOS bash 3.2 compat in release script: replace cp --parents,
local -n (nameref), and mapfile with POSIX-safe alternatives
- Fix _TOML_AGENTS: remove qwen (uses markdown per release script)
- Rename --from-github to --offline (opt-in to bundled assets)
- Add _locate_release_script() for cross-platform script discovery
- Update tests: remove bash 4+/GNU coreutils requirements, handle
Kimi directory-per-skill layout, 576 tests passing
- Update CHANGELOG and docs/installation.md
* Potential fix for pull request finding
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
* fix(offline): error out if --offline fails instead of falling back to network
- _locate_core_pack() docstring now accurately describes that it only
finds wheel-bundled core_pack/; source-checkout fallback lives in callers
- init() --offline + no bundled assets now exits with a clear error
(previously printed a warning and silently fell back to GitHub download)
- init() scaffold failure under --offline now exits with an error
instead of retrying via download_and_extract_template
Addresses reviewer comment: https://github.com/github/spec-kit/pull/1803
* fix(offline): address PR review comments
- fix(shell): harden validate_subset against glob injection in case patterns
- fix(shell): make GENRELEASES_DIR overridable via env var for test isolation
- fix(cli): probe pwsh then powershell on Windows instead of hardcoding pwsh
- fix(cli): remove unreachable fallback branch when --offline fails
- fix(cli): improve --offline error message with common failure causes
- fix(release): move wheel build step after create-release-packages.sh
- fix(docs): add --offline to installation.md air-gapped example
- fix(tests): remove unused genreleases_dir param from _run_release_script
- fix(tests): rewrite parity test to run one agent at a time with isolated
temp dirs, preventing cross-agent interference from rm -rf
* fix(offline): address second round of review comments
- fix(shell): replace case-pattern membership with explicit loop + == check
for unambiguous glob-safety in validate_subset()
- fix(cli): require pwsh (PowerShell 7) only; drop powershell (PS5) fallback
since the bundled script uses #requires -Version 7.0
- fix(cli): add bash and zip preflight checks in scaffold_from_core_pack()
with clear error messages if either is missing
- fix(build): list individual template files in pyproject.toml force-include
to avoid duplicating templates/commands/ in the wheel
* fix(offline): address third round of review comments
- Add 120s timeout to subprocess.run in scaffold_from_core_pack to prevent
indefinite hangs during offline scaffolding
- Add test_pyproject_force_include_covers_all_templates to catch missing
template files in wheel bundling
- Tighten kiro alias test to assert specific scaffold path (download vs offline)
* fix(offline): address Copilot review round 4
- fix(offline): use handle_vscode_settings() merge for --here --offline
to prevent data loss on existing .vscode/settings.json
- fix(release): glob wheel filename in create-github-release.sh instead
of hardcoding version, preventing upload failures on version mismatch
- docs(release): add comment noting pyproject.toml version is synced by
release-trigger.yml before the tag is pushed
* fix(offline): address review round 5 + offline bundle ZIP
- fix(offline): pwsh-only, no powershell.exe fallback; clarify error message
- fix(offline): tighten _has_bundled to check scripts dir for source checkouts
- feat(release): build specify-bundle-v*.zip with all deps at release time
- feat(release): attach offline bundle ZIP to GitHub release assets
- docs: simplify air-gapped install to single ZIP download from releases
- docs: add Windows PowerShell 7+ (pwsh) requirement note
* fix(tests): session-scoped scaffold cache + timeout + dead code removal
- Add timeout=300 and returncode check to _run_release_script() to fail
fast with clear output on script hangs or failures
- Remove unused import specify_cli, _SOURCE_TEMPLATES, bundled_project fixture
- Add session-scoped scaffolded_sh/scaffolded_ps fixtures that scaffold
once per agent and reuse the output directory across all invariant tests
- Reduces test_core_pack_scaffold runtime from ~175s to ~51s (3.4x faster)
- Parity tests still scaffold independently for isolation
* fix(offline): remove wheel from release, update air-gapped docs to use pip download
* fix(tests): handle codex skills layout and iflow agent in scaffold tests
Codex now uses create_skills() with hyphenated separator (speckit-plan/SKILL.md)
instead of generate_commands(). Update _SKILL_AGENTS, _expected_ext, and
_list_command_files to handle both codex ('-') and kimi ('.') skill agents.
Also picks up iflow as a new testable agent automatically via AGENT_CONFIG.
* fix(offline): require wheel core_pack for --offline, remove source-checkout fallback
--offline now strictly requires _locate_core_pack() to find the wheel's
bundled core_pack/ directory. Source-checkout fallbacks are no longer
accepted at the init() level — if core_pack/ is missing, the CLI errors
out with a clear message pointing to the installation docs.
scaffold_from_core_pack() retains its internal source-checkout fallbacks
so parity tests can call it directly from a source checkout.
* fix(offline): remove stale [Unreleased] CHANGELOG section, scope httpx.Client to download path
- Remove entire [Unreleased] section — CHANGELOG is auto-generated at release
- Move httpx.Client into use_github branch with context manager so --offline
path doesn't allocate an unused network client
* fix(offline): remove dead --from-github flag, fix typer.Exit handling, add page templates validation
- Remove unused --from-github CLI option and docstring example
- Add (typer.Exit, SystemExit) re-raise before broad except Exception
to prevent duplicate error panel on offline scaffold failure
- Validate page templates directory exists in scaffold_from_core_pack()
to fail fast on incomplete wheel installs
- Fix ruff lint: remove unused shutil import, remove f-prefix on
strings without placeholders in test_core_pack_scaffold.py
* docs(offline): add v0.6.0 deprecation notice with rationale
- Help text: note bundled assets become default in v0.6.0
- Docstring: explain why GitHub download is being retired (no network
dependency, no proxy/firewall issues, guaranteed version match)
- Runtime nudge: when bundled assets are available but user takes the
GitHub download path, suggest --offline with rationale
- docs/installation.md: add deprecation notice with full rationale
* fix(offline): allow --offline in source checkouts, fix CHANGELOG truncation
- Simplify use_github logic: use_github = not offline (let
scaffold_from_core_pack handle fallback to source-checkout paths)
- Remove hard-fail when core_pack/ is absent — scaffold_from_core_pack
already falls back to repo-root templates/scripts/commands
- Fix truncated 'skill…' → 'skills' in CHANGELOG.md
* fix(offline): sandbox GENRELEASES_DIR and clean up on failure
- Pin GENRELEASES_DIR to temp dir in scaffold_from_core_pack() so a
user-exported value cannot redirect output or cause rm -rf outside
the sandbox
- Clean up partial project directory on --offline scaffold failure
(same behavior as the GitHub-download failure path)
* fix(tests): use shutil.which for bash discovery, add ps parity tests
- _find_bash() now tries shutil.which('bash') first so non-standard
install locations (Nix, custom CI images) are found
- Parametrize parity test over both 'sh' and 'ps' script types to
ensure PowerShell variant stays byte-for-byte identical to release
script output (353 scaffold tests, 810 total)
* fix(tests): parse pyproject.toml with tomllib, remove unused fixture
- Use tomllib to parse force-include keys from the actual TOML table
instead of raw substring search (avoids false positives)
- Remove unused source_template_stems fixture from
test_scaffold_command_dir_location
* fix: guard GENRELEASES_DIR against unsafe values, update docstring
- Add safety check in create-release-packages.sh: reject empty, '/',
'.', '..' values for GENRELEASES_DIR before rm -rf
- Strip trailing slash to avoid path surprises
- Update scaffold_from_core_pack() docstring to accurately describe
all failure modes (not just 'assets not found')
* fix: harden GENRELEASES_DIR guard, cache parity tests, safe iterdir
- Reject '..' path segments in GENRELEASES_DIR to prevent traversal
- Session-cache both scaffold and release-script results in parity
tests — runtime drops from ~74s to ~45s (40% faster)
- Guard cmd_dir.iterdir() in assertion message against missing dirs
* fix(tests): exclude YAML frontmatter source metadata from path rewrite check
The codex and kimi SKILL.md files have 'source: templates/commands/...'
in their YAML frontmatter — this is provenance metadata, not a runtime
path that needs rewriting. Strip frontmatter before checking for bare
scripts/ and templates/ paths.
* fix(offline): surface scaffold failure detail in error output
When --offline scaffold fails, look up the tracker's 'scaffold' step
detail and print it alongside the generic error message so users see
the specific root cause (e.g. missing zip/pwsh, script stderr).
---------
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
* feat(ai): add native support for Pi coding agent by pi+gpt 5.4
* docs(pi): document MCP limitations for Pi agent
* fix: unitended kimi agent mention added to update-agent-context.ps1
* fix: address reviewer feedback
* Apply suggestions from code review
Changes in AGENTS.md weren't part of my PR, but the Copilot feedback seems to be correct is correct. I've doublechecked it with contents of test_agent_config_consistency.py and create-release-packages scripts
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Add automated markdown linting to ensure consistent formatting across
all markdown files in the repository.
Changes:
- Add .markdownlint-cli2.jsonc configuration
- Create .github/workflows/lint.yml for CI/CD integration
- Fix all 908 existing markdown errors across 27 files
- Enforce ATX-style headings and asterisk emphasis
- Set consistent 2-space list indentation
This prevents markdown errors after project initialization and
maintains high documentation quality standards.
