| name | cross-repo-open-diff-reconciliation |
| description | Review recent open diffs across multiple repos, classify repeated release/docs/tests gaps, and route each gap to the right remediation skill. Use when several repos are carrying unreleased work and you need a deterministic audit before patching or curating new skills. |
Cross-Repo Open Diff Reconciliation
Use this skill to turn recent multi-repo work into a bounded remediation map.
Workflow
- Route the request through
$skill-hub.
- Run
$skills-cross-repo-radar with a bounded lookback window.
- Inspect open diffs and changed-file clusters for:
- release-hygiene misses
- docs-lockstep drift
- missing tests
- policy/contract-sensitive changes
- startup/runtime acceptance changes
- Classify each repo finding into a small remediation lane.
- Route findings to the most specific follow-on skill instead of patching blindly.
- Produce a deterministic reconciliation table with:
- repo
- gap category
- severity
- evidence paths
- suggested skill lane
- When the user wants new skill curation, cluster repeated findings across repos and propose new skills or upgrades only after the reconciliation map is complete.
Default Gap Routing
release_hygiene_missing -> $skill-arbiter-release-ops
docs_lockstep_missing -> $docs-alignment-lock
tests_missing -> $skill-common-sense-engineering
- startup/launcher drift ->
$desktop-startup-acceptance
- cross-repo contract/policy drift ->
$skill-enforcer
Required Evidence
- radar JSON output
- changed-file samples per repo
- category/severity counts
- remediation mapping for every flagged repo
Scope Boundary
Use this skill for recent-work reconciliation and remediation routing.
Do not use it for:
- deep runtime debugging after the gap is already classified
- single-repo PR implementation when a more specific skill already fits
- skill admission itself; use
$skill-auditor and $skill-arbiter-lockdown-admission after curation
References
references/finding-taxonomy.md
Loopback
If the reconciliation map is incomplete:
- capture the missing repo or evidence slice
- rerun the radar with narrowed scope or corrected repo list
- continue only after every repo has a deterministic finding disposition