Help us improve
Share bugs, ideas, or general feedback.
From code
Checks if cross-repo coordinator results are reusable by comparing git hashes of peer repos against stored hashes, avoiding redundant Sonnet agent launches before Phase 1.4.1 peer discovery.
npx claudepluginhub closedloop-ai/claude-plugins --plugin codeHow this skill is triggered — by the user, by Claude, or both
Slash command
/code:cross-repo-cacheThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Check whether prior cross-repo coordinator results are still valid, avoiding redundant Sonnet agent launches when peer repositories haven't changed.
Skips Phase 7 build validation if no code changes since Phase 5 passed, using git diff hash caching. Stamp cache after Phase 5 success, check before Phase 7.
Checks for cached plan-evaluation.json using file timestamps to skip redundant plan-evaluator launches in Phase 1.3. Returns EVAL_CACHE_HIT with values or EVAL_CACHE_MISS if stale.
Coordinates changes across multiple repositories by syncing skills, templates, agents, and harness policies via git-mediated workflows or specification-mediated patterns for multi-repo projects.
Share bugs, ideas, or general feedback.
Check whether prior cross-repo coordinator results are still valid, avoiding redundant Sonnet agent launches when peer repositories haven't changed.
Activate this skill at the start of Phase 1.4.1 (Discover peers), before launching @code:cross-repo-coordinator. If the cache is fresh, skip the coordinator and use cached results.
The scripts/ directory is relative to this skill's base directory (shown above as "Base directory for this skill").
bash <base_directory>/scripts/check_cross_repo_cache.sh <WORKDIR>
After the cross-repo-coordinator runs successfully, stamp the cache:
# Compute hash of peer repo states
if [ -f ".workspace-repos.json" ]; then
repo_hashes=""
for repo_path in $(python3 -c "import json; [print(r['path']) for r in json.load(open('.workspace-repos.json')) if r.get('path')]" 2>/dev/null); do
[ -d "$repo_path/.git" ] && repo_hashes="$repo_hashes$repo_path:$(git -C "$repo_path" rev-parse HEAD 2>/dev/null || echo unknown) "
done
echo "$repo_hashes" | shasum -a 256 > $WORKDIR/.cross-repo-hash
else
shasum -a 256 $WORKDIR/.cross-repo-needs.json > $WORKDIR/.cross-repo-hash
fi
CROSS_REPO_CACHE_HIT
status: NO_CROSS_REPO_NEEDED | CAPABILITIES_IDENTIFIED
capabilities:
- peer-name: capability description
Action:
NO_CROSS_REPO_NEEDED: Mark 1.4.x phases complete, proceed to Phase 2CAPABILITIES_IDENTIFIED: Skip coordinator, proceed to Phase 1.4.2 with cached capabilitiesCROSS_REPO_CACHE_MISS
reason: <why the cache is stale or missing>
Action: Launch @code:cross-repo-coordinator as normal. After it completes, stamp the cache.
The script hashes the git HEAD commit of each peer repository listed in .workspace-repos.json. If any peer has new commits, the hash changes and the coordinator re-runs. This correctly handles the case where a peer repo is updated between iterations.
$WORKDIR/.cross-repo-needs.json (written by coordinator)$WORKDIR/.cross-repo-hash (written by stamp step).learnings/cross-repo-cache/{repo-name}/ (used by generic-discovery agents)