triaging-module-requests

name	triaging-module-requests
description	Triage MultiQC `module: new` GitHub issues: calculate 0-100 priority scores, apply priority labels, post analysis comments with score breakdowns, and give contributors actionable feedback to improve their request. Use when a new `module: new` issue is opened, when a user comments `@claude analyze-module` on a request, during weekly bulk triage, or when manually re-evaluating a module request.

Triage MultiQC Module Requests

Invoked by .github/workflows/module-requests.yml on new module: new issues, on @claude analyze-module comments, and on a weekly schedule (Mondays 9 AM UTC). Also runs on manual workflow dispatch.

Workflow

1. Pick a mode from context:
   • analyze-single — one issue (new issue or on-demand request)
   • triage-all — every open module: new issue (weekly batch)
   • dry-run — calculate and print, but make no GitHub changes
2. Fetch issue data with gh issue view / gh issue list. See github-actions.md for the exact commands.
3. Fetch tool metrics via scripts/fetch-tool-metrics.js (GitHub stars, PyPI/Conda downloads, last commit date).
4. Calculate the priority score using the rubric in scoring-criteria.md.
5. Apply the priority label (see Priority bands below). Remove any existing priority labels first; see github-actions.md.
6. Post the analysis comment using analysis-templates.md. Always show the score breakdown, what's good, and concrete improvements (with +X points tags where they apply).

Priority bands

Score	Band	Label
≥70	🔴 High	module: prio-high
40–69	🟡 Medium	module: prio-medium
20–39	🟢 Low	module: prio-low
<20	⚪ Hold	module: prio-hold

The five score categories (full rubric in scoring-criteria.md):

• Tool Popularity (25) — GitHub stars + maintenance bonus
• Package Downloads (15) — PyPI / Conda / Bioconda monthly
• Community Engagement (35) — reactions, comments, duplicates
• Request Quality (20) — completed fields + example files
• Technical Feasibility (15) — output format, metric clarity, parsing

Feedback principles

Be specific (point to exact fields), be encouraging (acknowledge strengths first), be consistent (apply the rubric uniformly), be transparent (show the calculation). Cache API results when batch-processing to stay under rate limits.