Use the write_todos tool effectively for task planning and decomposition in Deep Agents. Use when users want to (1) implement task planning with write_todos, (2) break down complex tasks into subtasks, (3) track agent progress through todos, (4) debug why todos aren't completing, (5) design todo structures for different task types (research, coding, analysis), (6) understand todo status lifecycle and best practices, or (7) visualize todo progression from LangSmith traces.
Initialize, validate, and troubleshoot Deep Agents projects in Python or JavaScript using the `deepagents` package. Use when users need to create agents with built-in planning/filesystem/subagents, configure middleware/backends/checkpointing/HITL, migrate from `create_react_agent` or `create_agent`, scaffold projects with repo scripts, validate agent config files, and confirm compatibility with current LangChain/LangGraph/LangSmith docs.
Use this skill when you need to test or evaluate LangGraph/LangChain agents: writing unit or integration tests, generating test scaffolds, mocking LLM/tool behavior, running trajectory evaluation (match or LLM-as-judge), running LangSmith dataset evaluations, and comparing two agent versions with A/B-style offline analysis. Use it for Python and JavaScript/TypeScript workflows, evaluator design, experiment setup, regression gates, and debugging flaky/incorrect evaluation results.
Deploy and operate production agent servers with LangSmith Deployment. Use when work involves choosing Cloud vs Hybrid/Self-hosted-with-control-plane vs Standalone, preparing/validating langgraph.json, creating deployments or revisions, rolling back revisions, wiring CI/CD to control-plane APIs, configuring environment variables and secrets, setting monitoring/alerts/webhooks, or troubleshooting deployment/runtime/scaling issues for LangChain/LangGraph applications.
Fetch, organize, and analyze LangSmith traces for debugging and evaluation. Use when you need to: query traces/runs by project, metadata, status, or time window; download traces to JSON; organize outcomes into passed/failed/error buckets; analyze token/message/tool-call patterns; compare passed vs failed behavior; or investigate benchmark and production failures.
Implement LangGraph error handling with current v1 patterns. Use when users need to classify failures, add RetryPolicy for transient issues, build LLM recovery loops with Command routing, add human-in-the-loop with interrupt()/resume, handle ToolNode errors, or choose a safe strategy between retry, recovery, and escalation.
Design state schemas, implement reducers, configure persistence, and debug state issues for LangGraph applications. Use when users want to (1) design or define state schemas for LangGraph graphs, (2) implement reducer functions for state accumulation, (3) configure persistence with checkpointers (InMemorySaver/MemorySaver, SqliteSaver, PostgresSaver), (4) debug state update issues or unexpected state behavior, (5) migrate state schemas between versions, (6) validate state schema structure, (7) choose between TypedDict and MessagesState patterns, (8) implement custom reducers for lists, dicts, or sets, (9) use the Overwrite type to bypass reducers, (10) set up thread-based persistence for multi-turn conversations, or (11) inspect checkpoints for debugging.
Implement multi-agent coordination patterns (supervisor-subagent, router, orchestrator-worker, handoffs) for LangGraph applications. Use when users want to (1) implement multi-agent systems, (2) coordinate multiple specialized agents, (3) choose between coordination patterns, (4) set up supervisor-subagent workflows, (5) implement router-based agent selection, (6) create parallel orchestrator-worker patterns, (7) implement agent handoffs, (8) design state schemas for multi-agent systems, or (9) debug multi-agent coordination issues.