paper-relevance-filter

Installation
SKILL.md

paper-relevance-filter

Decide whether a fetched paper belongs in this week's digest. Output is a per-paper decision (KEEP / DROP / REVIEW), a 0-100 score, and a one-line rationale that the user can audit.

Workflow

- [ ] Step 1: Load relevance criteria + the watchlist + last-4-weeks kept-paper IDs
- [ ] Step 2: Score each paper on three axes (match, criteria, novelty)
- [ ] Step 3: Combine to a 0-100 score; map to KEEP / REVIEW / DROP via thresholds
- [ ] Step 4: Apply tie-breakers (cap output at requested max kept count)
- [ ] Step 5: Return decisions + a calibration summary

Step 1 — Inputs

Installs
14
Repository
lyndonkl/claude
GitHub Stars
128
First Seen
May 11, 2026
paper-relevance-filter — lyndonkl/claude