Open-source rwxmap labels API operations r/w/x with a 23MB local model replacing a paid LLM
Tight_Heron1730 · reddit · 2026-11-08
The open-source rwxmap tool reads OpenAPI specs and assigns each operation a risk letter: r (read-only), w (revertible change), x (irreversible — deletes, key issuance, user additions, external webhooks). Simple rules settle most operations; ambiguous ones previously went to a paid hosted model. As of 0.11.0, the author's rwxlite 0.3.1 replaces it: a 23MB model fine-tuned from all-MiniLM-L6-v2 on 102,700 labeled operations from 690 public providers, running at 17-20ms per operation on a laptop CPU with no per-call bill and no data leaving the machine. On 1,443 unseen operations from 7 vendors: 87.9% correct, 9.3% sent to human review, 2.8% too loose. Confidence threshold is tunable. Both tools are Apache 2.0 on npm.
More from coding & agent
- DataTalksClub frees course materials for Production AI Agents for Generative Media, projects application-only — Al_Grigor · 2026-11-08
- Engineer's data: 11x more merged PRs per month at AI-native Tano vs pre-AI PostHog — bibryam · 2026-11-08
- No Chinese models allowed: what's left for local document OCR when contracts rule out PaddleOCR and Qwen? — aidenclarke_12 · 2026-11-08
- ADLC means two different things: Agentic Development Lifecycle vs Agent Development Lifecycle — Thirumalaivasan_GJ · 2026-11-08
- Anthropic's Official claude-quickstarts Repo Nears 20k Stars for Claude API Developers — anthropics · 2026-11-08
- Working With AI: Getting Better Results by Being 'Dumber', Not Smarter — glcst · 2026-11-08