observability
Use when adding logging to services, setting up monitoring, creating alerts, debugging production issues, designing SLIs/SLOs, or implementing structured logging (Pino, Winston), metrics (Prometheus, DataDog, CloudWatch), or distributed tracing (OpenTelemetry).
api-design
Use when designing new REST APIs, reviewing API designs, establishing API standards, designing request/response formats, pagination, versioning, authentication flows, or creating OpenAPI specifications.
database-design
Use when designing database schemas, reviewing data models, planning migrations, optimizing slow queries, or establishing database patterns. Covers relational (PostgreSQL, MySQL, SQLite), document (MongoDB), and ORM-integrated (Prisma, Drizzle, TypeORM) projects.
deep-research
Use for multi-day technology evaluations, competitive analysis, architectural exploration, best practices research, or major strategic decisions requiring comprehensive investigation and stakeholder reports. Unlike spike (time-boxed hours), supports extended multi-day investigation.
documentation
Use when creating READMEs, API docs, ADRs, or user guides. Provides templates and patterns for each documentation type. Dispatched by /work for docs-type tasks.
error-handling
Use when designing error hierarchies, implementing React error boundaries, adding retry logic or fallbacks, creating API error responses, integrating error tracking (Sentry), or improving user-facing error communication.
feature-audit
Use when verifying feature completeness against PRD requirements, identifying gaps between backend implementation and user-facing accessibility, generating remediation tasks, or auditing across frameworks (Next.js, React Router, TanStack, React Native, Expo).
finishing-branch
Use after a task, story, epic, worktree, or development branch is implementation-complete and freshly verified. Presents merge, PR, keep, or discard choices and records evidence before cleanup.
plan-revision
Use when revising existing implementation plans, analyzing impact of task changes, exploring modifications to project scope or dependencies, or updating task hierarchies with full impact visibility before execution.
planning
Use when analyzing PRD documents, concept briefs, or feature specs to create implementation plans, breaking work into epics/stories/tasks with dependencies and estimates, or generating PROJECT.md and kanban tracking.
sdk-development
Use when building TypeScript SDKs, extracting shared code into packages, creating developer tooling libraries, designing clean API surfaces, or publishing to npm (public or private). Covers typed clients, error handling, multi-target bundling (ESM/CJS/browser).
security-audit
Use when reviewing code security, auditing dependencies for CVEs, checking configuration or secret security, assessing authentication and authorization patterns, identifying OWASP vulnerabilities (injection, XSS, CSRF), or addressing security concerns about implementations.
session-review
Use after completing work sessions to analyze agent behavior patterns, prepare session handoffs for continuity, document completed work, identify blockers, or preserve context for the next session.
spike
Use for time-boxed technical investigations — feasibility studies, architecture exploration, integration assessment, performance analysis, or risk evaluation.
systematic-debugging
Use when diagnosing bugs, failing tests, build failures, unexpected behavior, flaky tests, regressions, or production incidents. Requires reproduction, root cause evidence, one-hypothesis fixes, and regression verification before editing.
testing-strategy
Use when designing test architecture, building API test suites, validating API contracts, setting up component or E2E testing, managing test data, debugging flaky tests, reviewing coverage strategy, or organizing test files. Covers test pyramid, mocking (MSW), frontend (React Testing Library, Playwright), and CI integration.
verification-before-completion
Use before claiming work is done, fixed, passing, ready to commit, ready to PR, or ready to mark complete. Requires fresh verification evidence and explicit command output before success claims.
work-session
Use when starting AI development sessions, resuming interrupted work, managing multi-session projects, or orchestrating work with human checkpoint control (supervised, semi-auto, auto, or unattended modes).
worktrees
Use when managing isolated git worktrees for task development. Handles worktree creation by task type, story-based reuse, dependency installation, and cleanup. Integrated into /work workflow.
api-integration
Use when consuming external APIs, integrating third-party services, generating type-safe API clients, implementing authentication flows, or working with OpenAPI/Swagger, GraphQL, or REST specs. TypeScript-primary with language-agnostic patterns.
architecture-review
Use when auditing project structure, planning refactors, improving code organization, analyzing dependencies and module boundaries, or identifying structural issues. TypeScript/JavaScript-primary with language-agnostic patterns.
browser-verification
Use when a completed task changed UI-affecting files and browser verification is needed during /work. Checks visual elements, interactions, and console errors in a real browser via Playwright. Not a standalone skill.
ci-cd
Use when creating or debugging CI/CD pipelines, implementing deployment strategies (blue-green, canary, rolling), optimizing build times, reviewing pipeline security, or working with GitHub Actions, GitLab CI, CircleCI, Azure DevOps, or Bitbucket Pipelines.
cloud-infrastructure
Use when provisioning cloud resources, choosing between AWS services (Lambda vs ECS vs EKS), writing CDK/IaC, designing serverless or container architectures, configuring IAM/security groups, or optimizing cloud costs.