| name | MPy Issue Triage |
| description | This skill should be used when the user wants to find duplicate or related MicroPython issues/PRs, triage an issue for similarity to existing items, detect spam or off-topic issues, or check if a new issue has already been reported. Invoke when user mentions finding duplicates, triaging issues, checking for related PRs, or detecting spam. |
MPy Issue Triage
Detect duplicate, related, and off-topic issues/PRs across MicroPython repositories using semantic search and LLM assessment.
When to Use
- Finding duplicates of a specific issue or PR
- Triaging a new issue to see if it's already been reported
- Checking if a PR overlaps with existing work
- Detecting spam or off-topic issues
- Looking for related issues that might inform a fix
Workflow
Triage an Issue
uv run --project ${CLAUDE_PLUGIN_ROOT} mpy-triage issue <NUMBER> --repo micropython/micropython
Triage a PR
uv run --project ${CLAUDE_PLUGIN_ROOT} mpy-triage pr <NUMBER> --repo micropython/micropython
Options
--skip-summarize - Skip Haiku summarization, use only static fields for embedding
--skip-assess - Skip Sonnet assessment, return ranked candidates with scores only
--json - Output results as JSON for machine processing
--repo REPO - Target repository (default: micropython/micropython)
Data Management
Before first use, the database must be populated:
uv run --project ${CLAUDE_PLUGIN_ROOT} mpy-triage collect
uv run --project ${CLAUDE_PLUGIN_ROOT} mpy-triage summarize
uv run --project ${CLAUDE_PLUGIN_ROOT} mpy-triage assemble
uv run --project ${CLAUDE_PLUGIN_ROOT} mpy-triage embed
Check database status:
uv run --project ${CLAUDE_PLUGIN_ROOT} mpy-triage stats
Interpreting Results
Each candidate shows:
- Classification: DUPLICATE, LIKELY_DUPLICATE, RELATED, OFF_TOPIC, or UNRELATED
- Confidence: high, medium, or low
- Reasoning: Why this classification was chosen
- Suggested action: What the maintainer should do
Results include full GitHub URLs for easy navigation.