Agent
Use when executing a single autoresearch experiment iteration - implements code changes, runs benchmark, evaluates metrics, logs results to JSONL, and manages git state (commit or revert). Returns structured result to the orchestrator.
Agent
Use when performing a structured codebase review from a blank slate -- no existing plan or prior context. Receives a structural inventory and sampled file contents, analyzes across seven dimensions, and returns severity-classified findings with holistic remediation. Opus-powered multi-persona staff engineer panel.
Agent
Implements individual tasks from plans with TDD, skill application, verification, and git commits. Use when executing a specific task that requires writing, modifying, or testing code as part of a larger plan.
Agent
Use when investigating codebase structure for planning, evaluating how proposed changes interact with existing modules and external services, or assessing cross-boundary data flow and consistency implications.
Agent
Use when evaluating implementation approaches for operational impact, deployability, observability, failure modes, rollback safety, or maintainability during planning.
Agent
Use when evaluating how proposed changes fit into long-term system evolution, assessing flexibility for future requirements, or identifying scope and extensibility tradeoffs during planning.
Agent
Use when evaluating implementation approaches for security implications, threat surface changes, input validation boundaries, authentication and authorization patterns, or potential attack path escalation.
Agent
Use when evaluating implementation approaches for performance characteristics, resource usage, memory allocation patterns, cache behavior, or hot path efficiency during planning.