fix(junie): stop $ARGUMENTS/$speckit-... becoming required Junie arguments

[chelsealong]

Fixes #4794

Junie treats every $identifier in a command body as a required named argument. The installed commands contained $ARGUMENTS and the prose $speckit-..., so Junie silently refused to run them.

Junie's post_process_command_content now:

• rewrites $ARGUMENTS to Junie's free-text $prompt, and adds allowPromptArgument: true to the frontmatter (per the Junie custom-slash-commands docs);
• drops the or \$speckit-...`` aside.

Tests: test_junie_argument_placeholders_adapted and test_junie_setup_leaves_no_stray_dollar_arguments. They fail without the fix (2 failed, 32 passed on the junie test file with the source reverted). With the fix, pytest tests/integrations gives 2677 passed, 5 skipped.

AI disclosure: this change was written with Claude Code (AI-assisted).

[Copilot]

Copilot review overview

🟡 Changes recommended

Existing or incidental allowPromptArgument: text can prevent the required frontmatter setting from being enabled.

Review effort: Balanced
Findings: 1

Open (1)

• Check allowPromptArgument value before rewriting $ARGUMENTS · New

What changed in this PR

Adapts generated Junie commands to accept free-text arguments correctly.

Changes:

• Rewrites $ARGUMENTS to $prompt and removes misleading $speckit-... prose.
• Adds regression and generated-command coverage.

File	Description
src/​specify_cli/​integrations/​junie/​__init__.py	Adds Junie-specific argument adaptation.
tests/​integrations/​test_integration_junie.py	Tests placeholder rewriting and generated commands.

---

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[chelsealong]

Fixed in the latest commit (AI-assisted, Claude Code): _adapt_argument_placeholders now only inspects the frontmatter block and sets allowPromptArgument to true (replacing an existing value such as false, or inserting it), so body text can no longer suppress it. Added test_junie_allow_prompt_argument_false_is_enabled (fails on the previous code; tests/integrations: 2678 passed, 5 skipped).

[Copilot]

Copilot review overview

🟢 Approval recommended

The focused implementation addresses both reported blockers with positive and regression coverage.

Review effort: Balanced
Findings: None

Resolved since last review (1)

• Check allowPromptArgument value before rewriting $ARGUMENTS

[mnriem]

Please fix test & lint errors

[chelsealong]

AI-assisted (Claude Code): I couldn't reproduce a test or lint failure on the current head. ruff@0.15.0 check src tests passes, pytest tests/integrations gives 2678 passed, 5 skipped, and the rest of tests gives 6321 passed, 10 skipped (Linux, Python 3.x). The only red check is pytest (macos-latest, 3.13), and its log isn't readable from the fork. If a maintainer can paste the failing test name or traceback, I'll fix it. I've made no code changes.

[chelsealong]

AI-assisted (Claude Code): the macOS log is readable now. The single failure is tests/specify_cli/workflows/test_catalog_versions.py::test_exact_add_uses_historical_url_digest_and_requirements (workflow catalog digest check, exit code 1). It's in code this PR doesn't touch (the PR only changes the Junie integration and its test file), and the other 9187 tests pass, including all Junie tests. I made no code changes; a re-run of that job should show whether it's environment-specific.

[Copilot]

Copilot review overview

🟡 Changes recommended

The unrelated workflow archive changes violate repository scope discipline and should be split out.

Review effort: Balanced
Findings: 1

Open (1)

• Move unrelated workflow archive changes to a separate PR · New