WeaveRx

Medical AI GitHub issue triage with auditable drafts, local safeguards, and human-in-the-loop defaults.

WeaveRx classifies MONAI, nnU-Net, and related project issues — reproducibility blockers, dataset access, subgroup performance, privacy/DICOM, and clinical validation requests — and produces a review-ready draft with sources and safeguard scores (local heuristics, no extra LLM calls).

Quick links

• README on GitHub — install, screenshots, quickstart
• Configuration — environment variables
• Examples — JSON samples and CLI captures
• Ethics — not for clinical use; human review required

Install: pip install weaverx

See it in action

Command: weaverx triage --repo Project-MONAI/MONAI --issue 42 --mock

Safeguard checks are advisory — they flag drafts for review; they never auto-block posting.

Text fallbacks: examples/captures/

GitHub Action

Dry-run triage on new issues via action.yml:

- uses: FratresMedAI/WeaveRx@v0.1.2
  with:
    repo: ${{ github.repository }}
    issue_number: ${{ github.event.issue.number }}
    dry_run: "true"
  env:
    XAI_API_KEY: ${{ secrets.XAI_API_KEY }}

Documentation map

Topic	Page
Medical AI categories	reference/categories.md
Draft safeguards	reference/safeguards.md
CLI flags and JSON	reference/cli.md
LLM providers	reference/llm-providers.md
Releasing	releasing.md

Local preview: pip install -e ".[docs]" && mkdocs serve — see Documentation setup.