Agent Automation & Development Skills

Use when building multi-turn agents, tool-calling systems, agent orchestration, or autonomous workflows — covers loops, tool calling, branching, reflection patterns.

Use when asked to check whether review/audit recommendations were actually implemented, whether gaps were closed, or to re-score the repo — verifies claims against repo state instead of trusting status reports.

Use when creating a branch, naming it, deciding whether to use a worktree, or handling secrets accidentally committed to history — branch naming convention, trunk protection, and secrets-removal procedure.

Use when reviewing REST or HTTP API endpoints, controllers, or route handlers. Covers HTTP status codes, idempotency, versioning, auth, pagination, error shapes, and rate limiting. Load instead of the general code-review skill for API-focused reviews.

Use when reviewing code that touches the database or persistence layer. Covers index strategy, query count, N+1, transaction scope, migration safety, and connection pooling. Load instead of the general code-review skill for DB-focused reviews.

Use when reviewing frontend, UI, or component code. Covers accessibility (WCAG AA), state management, bundle size, keyboard navigation, hydration, and rendering performance. Load instead of the general code-review skill for UI-focused reviews.

Creates well-structured git commits following conventional commit format. Use when committing changes, preparing commits, or when asked to commit code.

Use when designing a database schema, writing migrations, reviewing SQL queries, or choosing between relational and document models — covers naming, index strategy, migration safety, N+1 prevention, and transaction boundaries.

Use when writing a Dockerfile, docker-compose file, or CI containerization config — multi-stage builds, layer caching, security (non-root user, minimal base images, no secrets in layers), and health checks.

Use before creating any new file or directory in an established project — covers guarded root paths, docs/, src/, conf/, the allowed-additions escape hatch, and how to ask for permission. Load this skill at the start of every session in a project that has .agentharness-guarded-paths.json.

Triggers on harness friction events (hook failures, ambiguous guidance, violated mandates, mismatched reality) encountered during any session. Agent must: address the immediate problem, log it locally to docs/operational/harness-feedback.md, and file it upstream to andr-ca/agentharness. Default-on; skip upstream filing only if .agentharness-no-upstream-feedback exists at repo root.

Use when asked to port, migrate, or add equivalent agent instructions, skills, or custom sub-agents for a different coding tool than the one already configured — e.g. "add Cursor support from our CLAUDE.md", "we're switching from Cursor to Codex, port our rules", "make this work for Copilot too", "port our review subagent to Codex". Covers both agentharness-linked projects (use the real generators, don't hand-write) and plain projects with hand-authored config (port by hand, same principles).