feat(mcp): improve tool descriptions for standalone LLM use#472
Conversation
webpack is a framework, not a subject-area domain. The scenario table and example 2 both incorrectly placed it in domains.
The MCP tool descriptions were written for humans who already understand cq, but LLMs see them cold without the skill protocol. Rewrite descriptions to convey when and why to call each tool, not just what it does, so LLMs can self-direct without the cq skill/plugin.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: ASSERTIVE Plan: Pro Run ID: 📒 Files selected for processing (1)
WalkthroughThe MCP tool descriptions for confirm, flag, query, status, and propose were rewritten with expanded parameter text. The cq skill prompts were updated so the webpack example tags place webpack under frameworks rather than domains. ChangesMCP tool descriptions and tag guidance
Possibly related PRs
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
There was a problem hiding this comment.
Pull request overview
This PR improves the MCP tool descriptions in the CLI MCP server so standalone LLM clients can better self-direct when to call query, propose, confirm, flag, and status. It also fixes SKILL.md examples to categorize webpack as a framework (not a domain), propagated across the synced prompt copies.
Changes:
- Expanded/clarified MCP tool descriptions for
query,propose,confirm,flag, andstatusto guide correct call timing and intent. - Updated SKILL.md scenario table + webpack example to move
"webpack"fromdomainsintoframeworks. - Minor schema description improvements for tool arguments (
domains,languages,frameworks,pattern,limit).
Reviewed changes
Copilot reviewed 8 out of 8 changed files in this pull request and generated 1 comment.
Show a summary per file
| File | Description |
|---|---|
| sdk/python/src/cq/prompts/SKILL.md | Fixes webpack classification in domains/frameworks examples. |
| sdk/go/prompts/SKILL.md | Same SKILL.md example updates for Go prompt copy. |
| plugins/cq/skills/cq/SKILL.md | Same SKILL.md example updates for plugin prompt copy. |
| cli/mcpserver/status.go | Updates status tool description (note: needs a small accuracy tweak). |
| cli/mcpserver/query.go | Rewrites query tool description + improves parameter descriptions. |
| cli/mcpserver/propose.go | Rewrites propose tool description + improves parameter descriptions. |
| cli/mcpserver/flag.go | Rewrites flag tool description + improves parameter descriptions. |
| cli/mcpserver/confirm.go | Rewrites confirm tool description + improves parameter descriptions. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
The rewritten status tool description dropped the qualifier that confidence distribution excludes the public commons. Restore it so standalone LLM consumers understand what the stats cover.
Summary
self-direct without the cq skill/plugin installed.
examples (scenario table and example 2), propagated via
make sync-prompts.Test plan
make lint-clipassesmake test-clipassesSummary by CodeRabbit