Development

Detect when an AI-authored PR is reverted and trigger a Reflection session for RCA. Monitors revert commits, matches to AI PRs, and initiates analysis. Scheduled to run hourly. Invoke with /revert-rca-loop.

Track loop metrics, trends, improvements. Invoke: /progress-tracker.

This skill should be used when the user asks to "create a migration", "run prisma migrate", "deploy database changes", "baseline a database", "set up Prisma CI/CD", or mentions database schema deployment. Provides Prisma Migrate best practices for development and production.

Check bundle size impact of current changes against size-limit baselines. Use when editing apps/* or packages/rialto source, before committing, or when user asks about bundle size.

Daily optimization loop for the implement-queue workflow. Measures queue efficiency via the queueEfficiency sensor, appends a trend point, logs the run, and files de-duplicated ready issues on regression. Async eval is scheduled separately — never run synchronously here. Invoke with /optimize-implement-queue.

Scaffold a new Fastify + Prisma backend service in the mattbutlerengineering monorepo. Creates the service directory, package.json, app bootstrap, Prisma schema, health route, tests, and updates Turborepo config.

Scaffold a new Fastify route in services/{reservations,users,agent} matching the house pattern — schema validation, auth, error envelope per ADR-002, SSE broadcast (if reservations), tests

Scaffold a Playwright E2E test in one of the apps that has a Playwright config, matching the existing test fixtures and auth patterns

Scaffold a new rialto design system component with all required files (component, CSS module, test, story, barrel export) following established conventions

Scaffold a new Architecture Decision Record in docs/adr/ with the repo's canonical format and the next available sequential number

Audit human-facing markdown (README, CONTRIBUTING, SECURITY, docs/**, package READMEs) for claims the repo no longer satisfies — commands that fail, flags that don't exist, described behaviour the code dropped. Runs after scripts/audit-markdown.mjs has settled the mechanical findings. Use weekly from the docs-audit routine, or after a refactor that moved user-facing surfaces.

Run the same lint + typecheck + architecture-audit + drift checks CI runs, locally and in parallel. Catches workspace-package issues (missing deps, prop drift, lint rule violations) and stale generated artifacts (dep-graph, llms.txt) before pushing — the failures CI would surface in 5 minutes show up in 30 seconds. Use before opening or pushing to a PR.

Sensor-driven continuous improvement loop. Collects metrics from all sensors, detects regressions, creates issues, verifies past fixes, and self-tunes thresholds. Invoke with /learning-loop.

Pick oldest ready, implement via mbe agent run, create PR. ready→in-progress→has-pr. Invoke: /issue-worker.

Drain the ready-issue backlog with quality: claim a batch of independent issues, implement each in parallel TDD worktree subagents, then serially merge green PRs. Replaces ship-loop. Use when the user says 'work the queue', 'drain the backlog', 'implement the ready issues', or invokes /implement-queue (optionally via /loop).

Autonomous feature-ideation cycle. Advances the current ideation batch (honor vetoes, decompose proposals past the veto window, sweep completed tracking issues) and, only when the previous batch is fully complete, generates a new batch of 4-5 feature-proposal issues grounded in PRODUCT.md and repo-committed signals. Invoke with /ideate. Designed to run daily inside the mbe-morning routine.

Use for any question about a codebase, its architecture, file relationships, or project content — especially when graphify-out/ exists, where the question should be treated as a graphify query first. Turns any input (code, docs, papers, images, videos) into a persistent knowledge graph with god nodes, community detection, and query/path/explain tools.

Check deploy status, trigger deploys, and debug deploy failures for the mattbutlerengineering monorepo. Covers static sites (Cloudflare Workers), API services (DigitalOcean App Platform), and infrastructure (Pulumi).

Batch-apply pnpm.overrides for CVE fixes in a single PR. Use when multiple Dependabot/security alerts need dependency overrides — collapses N identical workflows into one operation.

Create ready-labeled GitHub issues from a feature description so /implement-queue can drain them — analyzes the codebase and files a dependency-ordered chain of child issues plus a tracking issue. Produces GitHub issues, not a design document. Invoke with /decompose.

Configure and use OpenAI Codex CLI. Use when user mentions codex, wants to set up Codex, or asks about Codex config (.codex/config.toml). Trigger: /codex, codex config, set up codex, .codex/config.toml

Audit Claude Code instruction files (CLAUDE.md, AGENTS.md, .claude/rules/*) for staleness and gaps — dangling paths, skill/agent tables that drift from disk, mandates pointing at things that don't resolve, and repo surfaces no instruction file documents. Proposes fixes; never writes silently. Use when instructions feel out of date, after a big refactor or extraction, or from the monthly mbe-monthly-meta-audit routine.

Audit the repo's automation layer — .claude/hooks, .claude/agents, .claude/skills, and .github/workflows — for guards that exist but never fire, guards that fire wrongly, and recurring failures with no guard at all. Recommends new hooks/subagents/skills grounded in real repo history. Proposes; never writes silently. Use when a known failure keeps recurring, after adding a hook or agent, or from the monthly mbe-monthly-meta-audit routine.

Seed non-breaking but detectable bugs (lint violations, dead links) to verify autonomous audit/lint loops catch and file issues. Scheduled to run weekly to test measurement machinery. Invoke with /chaos-agent or schedule via RemoteTrigger.