What problem does it solve?
Maintaining a security-critical open-source project like vm2 means handling a continuous stream of GitHub Security Advisory reports, and without a disciplined process reports pile up unacknowledged, duplicates get patched twice, reporters lose credit, and embargoed details risk leaking into public git history.
Core Features & Use Cases
- Advisory triage and scope decisions: Sweeps the open advisory queue, applies a documented in-scope/out-of-scope test based on whether a documented defense fails its promise, and escalates grey-zone reports to the owner.
- Clustering and deduplication: Groups reports by root cause using the same-fix test (one patch closing both PoCs at the same chokepoint), assigns primary ownership by creation date, and preserves reporter credit across merged duplicates.
- Reporter correspondence and ledger tracking: Posts acceptance, closure rationales, and fix summaries to advisory threads, and records every decision in a gitignored ledger so state survives across sessions.
- Use Case: At the start of a maintenance session, ask the assistant to work the security queue: it sweeps open advisories via the GitHub API, confirms clusters by reading threads, accepts or closes reports with rationales, delegates patching to the fix-vulnerability skill, and hands off confirmed fixes for merge and publication.
Quick Start
Ask the assistant to triage the current vm2 security advisory queue and report which reports are duplicates, which need fixes, and which should be escalated.