plugins

Technical leader and system designer for planning architecture, evaluating trade-offs, and creating implementation plans. Use when the task requires high-level design, system planning, or technical decision-making before writing code.

Knowledgeable technical assistant for answering questions about the codebase, explaining concepts, and providing guidance — without making any changes. Use when the user wants to understand something rather than change something.

Proactive code reviewer that catches bugs, security issues, and style violations in recent changes. Use after code changes and before committing.

Skilled software engineer for writing, editing, and refactoring code. Use when the task involves implementing features, fixing code, or making direct code changes. This is the default workhorse agent.

Database specialist for writing and reviewing migrations, queries, schema design, and ORM code. Use for database schema changes, query optimization, and migration safety checks.

Expert diagnostician for systematic troubleshooting, bug investigation, and issue resolution. Use when the user reports a bug, error, or unexpected behavior that needs investigation.

DevOps and infrastructure specialist for Dockerfiles, CI/CD pipelines, deployment configs, and infrastructure-as-code. Use for Docker, GitHub Actions, Terraform, Kubernetes, and deployment tasks.

Documentation writer that generates and updates READMEs, API docs, inline comments, and changelogs. Use when documentation needs to be created or updated.

Strategic workflow coordinator that breaks down complex tasks and delegates them to specialized agents. Use for complex, multi-step tasks that span multiple concerns — e.g., "build a new feature end-to-end", "investigate this issue and fix it", or "add a feature with tests, docs, and a PR."

Crafts pull request titles, descriptions, and commit messages by analyzing diffs and commit history. Use when preparing code for review or merging.

Refactoring specialist that improves code structure without changing behavior. Use when code works but needs cleanup — duplication, complexity, dead code, or poor organization.

Deep research agent that investigates external docs, APIs, libraries, and best practices using web search. Use when you need to evaluate a library, understand an API, or research approaches before coding.

Security auditor that scans for vulnerabilities, reviews auth flows, checks dependencies, and identifies secrets in code. Use for security reviews and hardening. Read-only — does not modify code.

Test writing specialist that creates thorough unit and integration tests. Use when code needs test coverage — new features, untested code, or after bug fixes.