overcode

Fires a pre-crafted workflow prompt to chain aidd skills in one command, or rewrite a prompt file in place (smarten). Use when you want to trigger plan→challenge or implement→review sequences, project snapshot with status context (previously), compress a .md prompt to its minimal form (smarten), or reconcile a mockup vs implementation screenshot (mirror). Do NOT use for single-skill tasks, custom workflows, or when you need direct control over each step.

Audits an ActivityPub federation implementation against a stack-aware checklist (inbox idempotency, HTTP signature verification, fan-out delivery, actor/key caching, outbox pagination conformance, rate limiting, circuit breaker, AS2 conformance, security, observability) and produces a ranked roadmap. Detects the AP stack and loads stack-specific pivots from installed sc-* plugins (.claude/rules/07-quality/ap-pivots-*.md); falls back to a generic 11-section schema otherwise. Use when the user mentions federation, ActivityPub, fediverse, AP delivery, inbox, outbox, HTTP signatures, WebFinger, actor, fan-out, "AP lent", "inbox lent", "delivery fail", "signature invalid", or invokes /ap-optimize.

Behavioural-test harness for skills, agents, and prompt-driven workflows — scaffolds scenario suites (Situation → Expected behaviour → Pass criteria), runs a dry-run judge harness that scores each scenario against a populated fixture without mutating real data, and re-runs suites for non-regression. Triggers on "behave", "scaffold behavioural tests", "run the behavioural suite", "regression-test this skill/agent", "write a scenario suite for", or "/behave <action>". Do NOT use for unit/integration tests of code (use the project's test runner), for code review against a style guide (use review/foresee), or for one-off manual checks.

Generates or updates CHANGELOG.md from git history following the Keep a Changelog format, commits the changelog, and creates a signed annotated tag. Use when a user wants to release or document changes: "generate changelog", "update CHANGELOG", "release v1.2.0", "tag this version", "what changed since last release". Do NOT use for writing release notes in a format other than Keep a Changelog, managing GitHub Releases UI, or bumping version numbers in package.json or Cargo.toml — this skill only manages CHANGELOG.md and git tags.

Audits a project's data layer (client + server) against a stack-aware checklist (N+1, query count, pagination, real-time listeners, payload size, cache strategy, quota/cost, indexing, security rules, observability) and produces a ranked roadmap. Detects the data-layer stack and loads stack-specific pivots from installed `sc-*` plugins (`.claude/rules/07-quality/data-pivots-*.md`); falls back to a generic 12-section schema + REST vanilla pivots otherwise. Use when the user mentions data perf, API perf, slow query, data-layer N+1 (queries répétées sur même collection/table — distinct from web-optimize's render-time N+1), quota, "trop de reads", "trop de requêtes", egress, rate limit, cold start, connection pool, réplication, read replica, sharding, "API lente", "backend lent", "DB perf", "audit data", "audit backend", or invokes /data-optimize.