LangChain — every error, fix, and pattern

What is LangChain?

LangChain is the most widely-used open-source framework for building applications on top of LLMs. It provides composable primitives — Runnable, prompt templates, chat models, retrievers, agents, memory, and structured output — that plug into any supported provider (OpenAI, Anthropic, Google, Bedrock, and dozens more) through a uniform API.

This hub focuses on the errors, gotchas, and migration issues that come up in production LangChain code — not the tutorial content. Every guide has real error messages, ranked causes, three tested fixes with working code, and prevention checklists.

Browse by category

Every LangChain error we've documented, grouped into the six categories that matter for troubleshooting.

All 15 LangChain errors

The complete list — sorted by page number. Each fix guide includes real error messages, root causes ranked by frequency, three tested fixes with working Python code, a seven-item prevention checklist, and five FAQs.

• #91
  RunnableSequence input schema mismatch

  LCEL & Runnables
  n/a
• #92
  RunnableParallel dict shape confusion

  LCEL & Runnables
  n/a
• #93
  .stream() returning full output at once

  LCEL & Runnables
  n/a
• #94
  create_tool_calling_agent schema errors

  Agents & Tool Calling
  n/a
• #95
  AgentExecutor infinite loop (cost blowup)

  Agents & Tool Calling
  n/a
• #96
  bind_tools schema mismatch across providers

  Agents & Tool Calling
  n/a
• #97
  ReAct agent "Could not parse LLM output"

  Agents & Tool Calling
  n/a
• #98
  RunnableWithMessageHistory session_id and config errors

  Memory & Chat History
  n/a
• #99
  Deprecated memory classes migration

  Memory & Chat History
  n/a
• #100
  VectorStoreRetriever score, MMR, and filter failures

  Retrieval & RAG
  n/a
• #101
  Document splitter chunk_size and overlap

  Retrieval & RAG
  n/a
• #102
  with_structured_output Pydantic and method errors

  Structured Output & Packaging
  n/a
• #103
  Provider package version drift

  Structured Output & Packaging
  n/a
• #104
  LangSmith tracing not appearing in dashboard

  Observability & Deployment
  n/a
• #105
  LangServe deployment errors

  Observability & Deployment
  n/a

LangChain package catalog (2026)

LangChain is not one package — it is a family of independently versioned packages. Understanding which one your import comes from is essential for debugging and upgrades.

For version compatibility across the family, see the version drift page.

Core patterns to know

Six patterns that come up in almost every LangChain codebase — and where each one goes wrong.

• LCEL composition with the | operator. Shape errors between steps are the most common issue. See RunnableSequence input mismatch.
• Tool-calling agents with create_tool_calling_agent. Schema and iteration limits matter. See tool schema errors and infinite loops.
• Message history with RunnableWithMessageHistory. The modern replacement for legacy memory classes. See migration guide.
• RAG retrieval with VectorStoreRetriever and document splitters. Configuration and chunking dominate quality. See retriever failures.
• Structured output with with_structured_output. Provider-specific methods matter. See structured output errors.
• Streaming across the full chain. Every step must support streaming. See streaming not chunking.