Agent Claude Code
A system-architecture agent for planning new subsystems, major technology choices, and designs that span several modules. It records decisions as ADRs, or short documents explaining an architecture decision and its reasons.
Agent Claude Code
A build-error-fixing agent for compiler, TypeScript, import, dependency, and other build errors. TypeScript is a programming language that checks many code mistakes before the program runs.
Agent Claude Code
A code review agent that checks finished features or modules, and code before a Git commit, for security weaknesses, quality problems, and technical debt.
Agent Claude Code
A runtime bug investigator who finds the root cause of software behaving incorrectly while it runs. It first makes the problem reproducible, narrows down where it occurs, tests explanations, and writes a test that reproduces it before changing the code.
Agent Claude Code
A deployment and operations specialist for putting software into production and running it reliably. It covers automated build-and-release pipelines, containers, Kubernetes, infrastructure-as-code, zero-downtime releases, and production monitoring.
Agent Claude Code
A documentation specialist that builds a map of the codebase and maintains API documentation and the technical README from the source code. An API is the defined way that software components communicate.
Agent Claude Code
An end-to-end testing specialist that uses Playwright to check complete user journeys in a web interface. It also checks whether those journeys work across different browsers.
Agent Claude Code
A planning guide for large coding tasks and refactors. It creates a staged blueprint with clear checks for completion and saves it in the project's plans/ folder.
Agent Claude Code
A cleanup specialist for removing code that is no longer used, unused exports, duplicate code, and related dependencies. It uses tools such as knip, ts-prune, and depcheck to find cleanup candidates.
Agent Claude Code
A security reviewer for application code and infrastructure. It checks common web risks from the OWASP Top 10, exposed secrets, dependencies, and infrastructure settings.
Agent Claude Code
A guide to test-driven development (TDD), where you write a failing test, make it pass, and then improve the code. It applies different test-coverage targets to standard and critical tasks.
Agent Claude Code
A testing specialist who strengthens tests after code has been implemented. It reads test reports, adds coverage where the code is weak, and maintains the testing setup.
Agent Claude Code
A frontend UI builder for creating web pages and interface components. It follows the project's DESIGN.md rules and design tokens, which are shared values for things such as colors, spacing, and typography.
Agent Claude Code
A manager for templates used in product and engineering process documents, such as product requirements, architecture decisions, and design documents.
Command Claude Code
A command for recording architecture decision records, written explanations of important technical choices. It saves why one option was chosen over another in context/decisions/.
Command Claude Code
A command for viewing a subagent's activity history, reports, and handoff decisions. A subagent is a delegated worker that performs part of a larger coding task.
Command Claude Code
A command for gradually fixing build and type errors with small, safety-focused changes.
Command Claude Code
A command that evaluates the quality of an entire software project and recommends which review agent should handle it. It routes the work first rather than performing the review itself.
Command Claude Code
A deployment command that first runs a security check, then delegates the release work to a deployment specialist. CI/CD means automated build-and-release pipelines, while IaC means managing infrastructure through code.
Command Claude Code
A command for maintaining project dependencies, the external packages that your code relies on. It identifies the package ecosystem, groups updates, and handles patch, minor, and major upgrades in stages.
Command Claude Code
A documentation command that creates project specifications suited to different development stages: demo, MVP, or full product.
Command Claude Code
A command for creating and running end-to-end tests with Playwright, a browser testing tool. It can test complete user journeys and capture screenshots, videos, and trace files.
Command Claude Code
A command for coordinating delegated work between agents. It analyzes a task and suggests a suitable agent for it.
Command Claude Code
A command that examines the current coding session and extracts reusable patterns worth saving as knowledge.