[kevin9327]

Prerequisite

Depends on #5215, which adds the shared IntegrationBase.process_registered_command() hook. Please merge #5215 first; this branch contains it via a merge commit, so the diff here shows both until #5215 lands.

With #5215 in place, this PR removes the is_cline_ext branch and the _hyphenate_* helpers from agents.py and applies the rewrite in src/specify_cli/integrations/cline/ through ClineIntegration.process_registered_command(), still only for non-core commands. docs/integrations/cline.md documents the behavior.

Fixes #5097

Problem

For Cline extension/preset commands, _hyphenate_body_refs (and _hyphenate_frontmatter_refs) rewrite every speckit.<a>.<b> token. A file name such as speckit.git.commit.md matches as a reference, so .specify/extensions/git/commands/speckit.git.commit.md becomes .../speckit-git-commit-md (extension lost).

Fix

Both helpers now share one compiled pattern and a small replacer that leaves a match ending in .md unchanged. Command references such as /speckit.git.commit are still hyphenated.

Evidence

• Added test_cline_body_refs_keep_md_file_names and test_cline_frontmatter_refs_keep_md_file_names in tests/integrations/test_integration_cline.py.
• On main both fail; with this change both pass.
• pytest tests/integrations tests/test_extensions.py: 3632 passed, 8 skipped.

AI disclosure

Implemented by Grok Bot (an AI coding agent) acting on behalf of @kevin9327, running autonomously; model unknown/not exposed to the agent. Extent: code generation, tests, and this description. The commit carries an Assisted-by: trailer. I ran the tests above locally.

[Copilot]

🟡 Changes recommended

A moderate active issue remains because the .md suffix check misclassifies valid command references such as /speckit.foo.md, leaving required transformations incorrect in both body and frontmatter without a regression test.

1 open issue

• Handle .md command references separately from filenames · New

What changed in this PR

This pull request updates Cline reference rewriting so command filenames ending in .md retain their extension while dotted command references remain hyphenated.

Changes:

• Shares compiled matching and replacement logic for body and frontmatter references.
• Preserves .md-suffixed command filenames.
• Adds regression tests for body and frontmatter handling.

File	Summary
tests/​integrations/​test_integration_cline.py	Adds regression coverage for preserving .md command filenames in body and frontmatter references.
src/​specify_cli/​agents.py	Centralizes Cline reference rewriting, but the .md suffix check also leaves valid command references such as /speckit.foo.md unchanged; this requires a context-aware distinction.

_{Review effort: Lite (auto)}

[mnriem]

Please address Copilot feedback

[Copilot]

🟡 Changes recommended

An active moderate issue remains because dotted preset command names can be misclassified as filenames, producing incorrect Cline references without a regression test.

1 open issue

• Handle dotted preset command names ending in md · New

1 resolved since last review

• Handle .md command references separately from filenames

_{Review effort: Lite (auto)}

[mnriem]

Please address Copilot feedback

[Copilot]

🔵 Needs a closer look

An active moderate finding remains because bracketed slash-prefixed references are still transformed incorrectly, affecting body and frontmatter processing.

No open issues 🎉

1 previously missed issue

In code that hasn't changed since last review

Handle bracketed slash command references

src/​specify_cli/​agents.py:118

The slash-prefixed command rule still fails when the reference is immediately preceded by a Markdown [ (for example [/speckit.foo.md]): text[start - 2] is [ here, so this branch returns the dotted token unchanged instead of producing /speckit-foo-md. This can affect both body and recursively processed frontmatter strings; treat bracketed command delimiters as command context (or use an explicit path check) and add a regression case.

1 resolved since last review

• Handle dotted preset command names ending in md

_{Review effort: Lite (auto)}

[mnriem]

Please address Copilot feedback. Note that all of this should be really replaced by the author using SPECKIT_COMMAND_MYEXT_MYCMD style and we will eventually mandate that

[kevin9327]

Thanks, addressed the latest Copilot finding in d00721f. A speckit.*.md token is now kept as a file name only when the slash before it follows a path-name character (letter, digit, ., _, -, ~, /, \, }, $), so [/speckit.foo.md] and {/speckit.foo.md} are hyphenated like any other slash command, in the body and in frontmatter. Added test_cline_refs_hyphenate_bracketed_slash_command_named_md, which fails on the previous commit and passes now; tests/integrations: 3149 passed, 8 skipped.

Noted on the __SPECKIT_COMMAND_MYEXT_MYCMD__ placeholders. This PR only keeps the existing dotted-reference rewrite from breaking .md file names until that becomes the required form.

[Copilot]

🟡 Changes recommended

A moderate active issue remains because valid Windows root-level paths can still be rewritten incorrectly, dropping filename protection and lacking regression coverage for both slash styles.

1 open issue

• Preserve .md protection for Windows drive-root paths · New

_{Review effort: Lite (auto)}

[Copilot]

🔵 Needs a closer look

Human assessment is needed because the change alters user-visible Cline registration behavior and carries integration-documentation implications not covered by the published review.

No open issues 🎉

1 resolved since last review

• Preserve .md protection for Windows drive-root paths

_{Review effort: Lite (auto)}

[mnriem]

The Cline-only gate explains the test scope, but it also highlights an architectural issue: agents.py should not contain integration-specific branches. As described in design/integration.md under “Integration delivery boundary,” shared behavior belongs in shared infrastructure, while Cline-specific behavior belongs in src/specify_cli/integrations/cline/. Please move this rewriting into the Cline integration through the integration contract, preserving its current non-core-only scope, and add registration-level regression coverage rather than only testing the private helpers directly. This can remain one focused PR if the existing contract supports it; if a shared capability is needed first, please split that prerequisite into a separate PR. Please also update the Cline integration documentation.

AI disclosure: Drafted for @mnriem by GitHub Copilot using GPT-6.1 Sol in interactive mode; AI assistance covered code/design assessment and drafting this response.

[kevin9327]

Moved the rewrite into the Cline integration in ab5d147, with the shared prerequisite split out as #5215.

• feat(integrations): add process_registered_command hook with source_id #5215 adds IntegrationBase.process_registered_command(frontmatter, body, *, source_id). register_commands() calls it for every command after name injection and before $ARGUMENTS/__SPECKIT_COMMAND_*__ resolution and rendering. The default is a no-op, and it is tested on its own with existing integrations (opencode, claude skills layout, aliases).
• This PR merges that branch, then removes is_cline_ext and the _hyphenate_* helpers from agents.py. ClineIntegration.process_registered_command() applies the rewrite and returns core commands unchanged, so the scope stays non-core-only.
• Added registration-level tests that call CommandRegistrar.register_commands("cline", ...). The extension-command test fails on main (the output contains commands/speckit-git-commit-md) and passes here. A second test confirms a core command is not rewritten. The existing helper tests now target the Cline module.
• Updated docs/integrations/cline.md.

Tests: pytest -n 8 tests/integrations tests/test_post_process.py tests/test_extensions.py tests/test_registered_command_hook.py tests/test_presets.py gave 3790 passed and 8 skipped. uvx ruff@0.15.0 check src tests is clean. In the full tests run, 9 tests fail in this environment, all outside the touched code: Rich table truncation in test_installed_adapters.py list output and four shell-script parity tests. I did not investigate them further.

AI disclosure: posted on behalf of @kevin9327 by Grok Bot, an AI coding agent running autonomously (model unknown, not exposed to the agent). It did the code, tests and this comment. Each new commit carries an Assisted-by: trailer. The agent ran the tests; no personal human review is claimed.

[Copilot]

🟡 Changes recommended

The path-character whitelist still rewrites valid paths with parentheses, percent signs, or brackets, stripping file extensions in both body and frontmatter processing.

1 open issue

• Preserve valid path suffixes during character normalization · New

_{Review effort: Balanced (auto)}

[mnriem]

Please address Copilot feedback

[Copilot]

🟢 Approval recommended

No active review issues remain for the changes that preserve Markdown filenames while normalizing Cline command references.

No open issues 🎉

1 resolved since last review

• Preserve valid path suffixes during character normalization

_{Review effort: Balanced (auto)}

[mnriem]

Thank you!