AI Engineering

Behavioral Modes defines distinct AI operating modes (brainstorm, implement, debug, review, teach, ship, orchestrate) that adapt Claude's behavior based on task type. Use it to switch modes automatically or manually for more effective responses.

Claude Code Guide provides templates and best practices for configuring Claude Code, prompting the agent, and debugging issues. The guide includes configuration templates, thinking keywords, and debugging techniques.

Context Window Management provides strategies for managing LLM context windows, including summarization, trimming, routing, and avoiding context rot. Use it when hitting token limits, losing critical information mid-dialogue, or tuning long-context applications.

CrewAI Multi-Agent orchestrates teams of autonomous AI agents that collaborate on complex tasks through role-based delegation and sequential or hierarchical workflows. Use it when building multi-agent systems with memory, tools, and observability without LangChain dependencies.

Docs Search queries auto-generated codebase documentation to find function signatures, class definitions, API docs, and code comments. Use it before implementing changes or when asked to search docs or look up a function.

LangChain is a framework for building LLM-powered applications with agents, chains, and RAG. Use it for chatbots, question-answering systems, autonomous agents, and rapid prototyping across multiple providers.

LlamaIndex is a data framework for building LLM applications with RAG, specializing in document ingestion, indexing, and querying. Use it for document Q&A, chatbots, knowledge retrieval, and building RAG pipelines over private data.

Memory Search searches conversation history and semantic memory to recall previous discussions, decisions, and context. Use when asked to search memory, check history, or before starting work to recall prior decisions.

Herdr inspects and controls Herdr-managed terminal panes, tabs, workspaces, and coding agents through the herdr CLI. Use it when the user explicitly mentions Herdr or asks to control a Herdr session, and only when HERDR_ENV=1.

Add New Skill provides a structured template for creating cybersecurity-focused Claude Code skills aligned with MITRE ATT&CK and NIST CSF. Use it when documenting a new security capability for the agent.

Beads provides a durable, shared task-tracking workflow for repositories that use the bd CLI. Use it to find ready work, claim or close tasks, manage blockers, and preserve project context across sessions.

Beads provides a Dolt-backed graph issue tracker for AI agents to manage multi-session work with dependencies and persistent memory across conversation compaction. Use when work spans sessions, has blockers, or needs context recovery after compaction.