Which agents are available and when to use each. Match the agent’s tool access to the task requirements.
Need Agent Type Why Search/read code Explore Read-only, fast, cheap (haiku) Architecture design Plan Read-only, thoughtful analysis Run commands/git Bash Shell access only Multi-step implementation general-purpose Full tool access Claude Code questions claude-code-guide Docs and web access Security review sdlc:security-reviewer Vulnerability assessment, secure coding Code quality review sdlc:quality-enforcer Formatting, linting, code style Code simplification code-simplifier:code-simplifier Clarity, consistency, minimalism Architecture analysis refactor:architect Design review, optimization planning Best practices research adr:adr-researcher Codebase analysis + web research Deep code exploration feature-dev:code-explorer Trace execution paths, map architecture Code review feature-dev:code-reviewer Bugs, logic errors, conventions Source code chunk analysis swarm:rlm-code-analyzer Code-aware, scope context, analysis focus Data/CSV chunk analysis swarm:rlm-data-analyzer Column-aware, distributions, statistics JSON chunk analysis swarm:rlm-json-analyzer Schema-aware, structural patterns. Also used by JSONL Log Analyzer General chunk analysis swarm:rlm-chunk-analyzer Haiku, fast, structured JSON output Synthesize chunk findings swarm:rlm-synthesizer Sonnet, aggregation and deduplication
Key rule: Read-only agents (Explore, Plan) cannot edit or write files. Never assign them implementation work.
Always available without plugins.
Task ({ subagent_type: " Bash " , prompt: " Check git status and show recent commits " })
Tools: Bash onlyBest for: Git operations, command execution, system tasks
Task ({ subagent_type: " Explore " , prompt: " Find all API endpoints " , model: " haiku " })
Tools: All read-only (no Edit, Write, Task)Model: Haiku (fast, cheap)Best for: Codebase exploration, file searches, code understanding
Task ({ subagent_type: " Plan " , prompt: " Design an OAuth2 implementation plan " })
Tools: All read-onlyBest for: Architecture planning, implementation strategies
Task ({ subagent_type: " general-purpose " , prompt: " Implement caching for the user API " })
Tools: All tools (full access)Best for: Multi-step tasks, research + action combinations
Task ({ subagent_type: " claude-code-guide " , prompt: " How do I configure MCP servers? " })
Tools: Read-only + WebFetch + WebSearchBest for: Questions about Claude Code, Agent SDK, Anthropic API
Available when the corresponding plugins are installed. Plugin agents use the format {plugin}:{agent-name}.
Agent Focus sdlc:security-reviewerVulnerabilities, supply chain security, secure coding sdlc:quality-enforcerFormatting, linting, error handling, documentation sdlc:compliance-auditorFull SDLC compliance audit feature-dev:code-reviewerBugs, logic errors, security, code quality, conventions code-simplifier:code-simplifierClarity, consistency, minimalism refactor:architectArchitectural analysis, optimization planning superpowers:code-reviewerReview against plan and coding standards human-voice:voice-reviewerContent voice and tone
Agent Focus adr:adr-researcherCodebase analysis + web search for best practices feature-dev:code-explorerTrace execution paths, map architecture, dependencies feature-dev:code-architectDesign feature architectures based on existing patterns
Agent Focus refactor:architectArchitecture analysis and optimization planning refactor:codeImplementation of architectural optimizations refactor:testTest coverage analysis and test generation
Agent Focus sdlc:ci-architectCI pipeline design and GitHub Actions configuration
Agent Focus auto-harness:test-generatorComprehensive test definitions for components
Content-aware chunk analyzers — the Team Lead selects the analyst based on detected content type. In multi-file directory analysis, different analyst types run simultaneously.
Agent Focus swarm:rlm-code-analyzerSource code analysis with function/class scope context, analysis focus (Haiku) swarm:rlm-data-analyzerCSV/TSV analysis with column distributions and statistics (Haiku) swarm:rlm-json-analyzerJSON/JSONL analysis with schema patterns and field distributions (Haiku). Also used by the JSONL Log Analyzer skill for log-specific schema-aware analysis swarm:rlm-chunk-analyzerGeneral-purpose: logs, prose, config, markup (Haiku) swarm:rlm-synthesizerAggregate chunk findings into coherent reports (Sonnet)
Do NOT override analyst models. Leave the model parameter unset when spawning — agent definitions default to Haiku, which is correct for structured analysis tasks.
