An agent harness that turns you into a manager of autonomous digital engineers.
Agent Claude Code
Implements code changes in an isolated context. Use for all implementation work — each task from tasks.md gets dispatched to an implementer. Independent tasks can run in parallel across multiple implementers.
Agent Claude Code
Read-only codebase exploration. Use when you need to investigate code architecture, trace call chains, analyze module dependencies, or gather context for planning and design. Returns summarized findings, not raw file contents.
Agent Claude Code
Code review focused on performance. Dispatched by the code-review skill.
Agent Claude Code
Use for reviewing low-level systems diffs (storage engines, distributed protocols, lock/memory/transaction code, critical runtime paths) where a change adds or modifies control-flow branches — allow-paths, fallbacks, type guards, nullability checks, trust boundaries exported to other subsystems. Best when subtle…
Agent Claude Code
Code review focused on robustness. Dispatched by the code-review skill.
Agent Claude Code
Code review focused on spec compliance. Dispatched by the code-review skill.
Agent Claude Code
Code review focused on code standards and conventions. Dispatched by the code-review skill.
Agent Claude Code
Per-finding code-review validator. Dispatched in parallel (one per finding) after reviewers return. Reads the cited code fresh, attempts a concrete reproduction or cites an exact rule, and renders a verdict with a confidence score and an optional committable suggestion.
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: