A design agent that turns a requirements report into a system design and an architecture report. It records technology choices, trade-offs, module structure, dependencies, and architecture decisions, but does not write source code.
A code-review agent that examines changes for quality, maintainability, performance, and design issues, then writes a review report. It does not edit source code or perform security reviews.
A read-only reviewer for requirements, system designs, and implementation plans. It looks for hidden assumptions, unclear points, and missing details, then produces a review report.
An implementation and debugging agent that follows a plan, changes code, fixes bugs, and prepares the result for testing. It does not design tests, perform security reviews, or carry out general code-quality reviews.
A documentation agent that reads a codebase and writes documents suited to a requested audience, purpose, type, and level of detail. It does not edit source code.
An interviewing agent that asks questions and turns the answers into a structured requirements report. Requirements describe what a system should do, along with its constraints and quality needs.
A planning agent that combines requirement, design, test, and review reports into a step-by-step implementation plan. It breaks work into tasks, sets priorities and milestones, and identifies tasks that can run in parallel.
A project setup agent that turns supplied technology-stack and coding-convention information into two project rule files. It can research official style, framework, security, and testing guidance before generating or updating those files.
A security review agent that examines code for vulnerabilities and writes a security review report. It checks areas such as injection, cross-site scripting, access control, exposed secrets, and known dependency vulnerabilities.
A debugging investigation agent that examines errors, recent changes, and related code to identify likely root causes. It produces an analysis report but does not change code.
A testing agent that designs test plans, writes test code, runs tests, and records the results in a test report. It writes failing tests first as part of TDD, or test-driven development, where tests guide the implementation.
A parallel-worktree coding agent that implements planned changes, fixes bugs, and refactors code. It works in an isolated Git worktree so another agent can work on the main project at the same time.
A worktree-only debugging agent for investigating the root cause of a developer's problem without changing the code. A worktree is an isolated working copy of a Git repository.
A parallel-worktree testing agent that designs tests, writes test code, runs it, and records the results. A worktree is an isolated copy of a Git project used so parallel tasks do not overwrite each other.
★not rated 1 yesterdayA60 tokens
originalMIT
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: