sync-skill-cache

What it does and when to use it

Sync the Logic-Lens working-copy skills/ into the installed plugin cache so content-evals test the EDITED skill, not the last published one. ALWAYS run this after editing any skills/**/SKILL.md or guide/_shared file and BEFORE running content-evals — otherwise the eval silently grades stale content and every token is wasted. Also use to verify the cache is in sync ("is the cache fresh?", "did my edit take effect in evals?", "check skill cache drift").

Triggers from the description

How it works

1. Steps

---
name: sync-skill-cache
description: Sync the Logic-Lens working-copy skills/ into the installed plugin cache so content-evals test the EDITED skill, not the last published one. ALWAYS run this after editing any skills/**/SKILL.md or guide/_shared file and BEFORE running content-evals — otherwise the eval silently grades stale content and every token is wasted. Also use to verify the cache is in sync ("is the cache fresh?", "did my edit take effect in evals?", "check skill cache drift").
disable-model-invocation: true
---

# sync-skill-cache

content-evals invoke each skill through Claude Code's plugin mechanism, which loads from
`~/.claude/plugins/cache/logic-lens-marketplace/logic-lens/<version>/skills/` — **not** the
repo's `skills/`. Editing `skills/` changes nothing the eval sees until you copy it into the
cache. This is the single most expensive footgun in the iteration loop: skip it and you grade
the previous skill while believing you tested your change.

## Steps

1. **Sync + verify** from the repo root:
   ```bash
   bash .claude/skills/sync-skill-cache/scripts/sync-cache.sh
   ```
   The script reads the active version from `package.json`, rsyncs `skills/` into the matching
   cache dir (`--delete`, so removed files are removed in the cache too), then diffs the two to
   confirm they are byte-identical. It prints `OK …` on success.

2. **If it prints `DRIFT` or a missing-cache error:** do not run evals. The two common causes:
   - The cache dir for the current `package.json` version doesn't exist (plugin not installed at
     this version) — reinstall/refresh the plugin, or check you didn't just bump the version
     without refreshing the cache.
   - rsync was blocked — rerun and read the diff list it prints.

3. **Verify-only mode** (no copy) to answer "is the cache fresh?":
   ```bash
   bash .claude/skills/sync-skill-cache/scripts/sync-cache.sh --check
   ```

This skill is the mandatory gate between editing a skill and grading it. The `iterate-skill`
orchestrator calls it automatically; run it by hand whenever you edit a skill outside that loop.

Versions

The database has no other skills named sync-skill-cache.

Database snapshot: 13.11.2026 · package version 11b63646a82e (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

• 100run-iteration-eval
  Run the Logic-Lens content-eval pipeline for one iteration and produce a scored summary.json — use to measure a skill change. Wraps scripts/run-content-evals.sh (runner, costs tokens) and scripts/grade-iteration.py (grader, free, re-runnable). ALWAYS sync the plugin cache first.…
• 85plugin-cache-sync-usage
  Show plugin-cache-sync status and explain the sync workflow. Use when the user asks which installed plugins are stale against their local source, how to make a local plugin edit live in ordinary sessions, or how plugin-cache-sync works.
• 100iterate-skill
  Run the Logic-Lens skill-improvement loop end to end — baseline → diagnose failures → edit → sync cache → re-eval → verify net gain → iterate until clean. Use whenever the goal is to RAISE a skill's eval score or fix a failing eval mode: "improve logic-review", "the format…
• 100refresh-plugin-cache
  This skill should be used when the user asks to "refresh the plugin cache", "update plugin cache", "sync plugin cache", "update installed plugins", "check for plugin updates", "are my plugins up to date", "is the plugin cache fresh", or "verify plugin versions". Scoped to…
• 100skill-eval
  Measure whether a skill actually helps: build a `claude plugin eval` suite, run it with and without the skill under a required spend cap, and read the delta. Use when adding evals to a skill, converting evals.json, re-checking skills after a model upgrade, or deciding if a skill…
• 100skill-eval
  Runs and analyzes eval test cases for Claude skills. Use when asked to "test a skill", "check if this skill works", "run evals", "verify skill quality", or after creating/modifying a skill.