Document Description Getting Started Installation, configuration, common commands Slash Commands /bab:chat, /bab:review, and other slash commands Provider Setup API keys, AI providers, environment variables, model selection Plugin Authoring Manifest, directory layout, SDK, plugin installation Adapter Tutorial Plugin interface, parsing patterns, extending tools
Tool Type Description chat Simple Collaborative thinking partner for brainstorming and development discussion thinkdeep Workflow Multi-stage investigation and deep reasoning for complex problems debug Workflow Systematic debugging and root cause analysis analyze Workflow Comprehensive code analysis (architecture, performance, quality) codereview Workflow Systematic code review with expert validation planner Workflow Interactive task planning with branching and revision consensus Specialized Multi-model consensus through structured debate delegate Specialized Run prompts through CLI plugin adapters (Claude, Codex, Copilot, OpenCode) refactor Workflow Code smell detection, decomposition, modernization analysis secaudit Workflow Security audit with OWASP, compliance, and threat modeling testgen Workflow Test suite generation with edge case coverage docgen Workflow Code documentation generation with complexity analysis precommit Workflow Pre-commit validation for git changes tracer Workflow Code tracing for execution flow and dependency mapping challenge Specialized Critical thinking — forces scrutiny of claims list_models Utility Lists available models from providers and plugins version Utility Returns bab and runtime version information
Simple : Single-pass tools — send a prompt, get a response. Direct AI provider call.Workflow : Multi-step investigation tools. Claude performs structured research across multiple steps, then optionally sends findings to an AI model for expert validation.Specialized : Custom implementations with unique behavior (multi-model debate, CLI bridging, critical thinking).Utility : Information tools with no AI model calls.