langchain-ai/langchain

Contributor readiness

Maintainer signals

Maintainers respond in ~<1 days (median first response, measured from real issue sampling; 100% of sampled issues got a response).

This project has no valid contribution manifest yet, so maintainer preferences are unknown.

Local setup

• CI: _compile_integration_test.yml, _lint.yml, _refresh_model_profiles.yml, _release.yml, _test.yml, _test_pydantic.yml, _test_vcr.yml, auto-label-by-package.yml, block_fork_main_prs.yml, bump_uv_pin.yml, check_agents_sync.yml, check_diffs.yml, check_extras_sync.yml, check_release_deps.yml, check_versions.yml, close_unchecked_issues.yml, codspeed.yml, integration_tests.yml, pr_labeler.yml, pr_labeler_backfill.yml, pr_lint.yml, pr_lint_trailer.yml, refresh_model_profiles.yml, remove_waiting_on_author.yml, reopen_on_assignment.yml, require_issue_link.yml, tag-external-issues.yml

Links

• GitHub repository ↗
• Contributing guide ↗
• Code of conduct ↗

• #33787 Agent re-attempts original tool call after HumanInTheLoopMiddleware edit decision

  intermediate · assigned · bug · langchain · external

• #38351 create_agent: model_to_tools router can return "model" but path_map omits it -> KeyError('model')

  intermediate · unassigned · bug · langchain · external

• #38352 Bug: SummarizationMiddleware._find_safe_cutoff_point preserves orphan ToolMessages -> provider 400 on next call

  intermediate · unassigned · bug · langchain · external

• #36862 AzureChatOpenAI Reasoning Summary not extracted properly

  intermediate · unassigned · bug · openai · external

• #38159 `HumanInTheLoopMiddleware`: Support regex/glob patterns in `interrupt_on` tool name keys

  intermediate · unassigned · langchain · feature request · external

• #34382 Model calls inside middleware are leaking when streaming

  intermediate · unassigned · bug · core · langchain · openai

• #38398 langchain-anthropic: `cache_control` on a `tool_use` content block is silently dropped when a matching `tool_calls` item exists

  intermediate · unassigned · bug · anthropic · external

• #38651 :Feature: Support Anthropic mid-conversation system messages (MidConversationSystemBlockParam / role:system in messages) in ChatAnthropic

  intermediate · unassigned · feature request · anthropic · external

• #36310 Expose Deep Agent Harness as Composable LangGraph Primitives

  intermediate · unassigned · langchain · feature request · external

• #37093 HumanInTheLoopMiddleware executes rejected tool calls in LangGraph's ToolNode

  intermediate · unassigned · bug · langchain · external

• #38544 `ProviderToolSearchMiddleware` defers tools as flat top-level functions instead of wrapping them in a `namespace` — no token savings

  intermediate · unassigned · bug · langchain · openai · external

• #37972 Add deterministic middleware orchestration with priority, conditional activation, and safe parallel execution

  intermediate · unassigned · core · langchain · feature request · external

• #36411 bug(langchain): agent terminates on return_direct tool even when tool call fails

  intermediate · unassigned · external

• #34974 Human In the Loop Middleware throws error when the agent is invoked as .ainvoke()

  intermediate · unassigned · bug · core · langchain · external

• #38780 `_parse_input` silently drops required fields resolved via `validation_alias` in `args_schema`

  intermediate · unassigned · bug · core · external

• #38814 SSRF via recursive sitemap fetch in `SitemapLoader` — `restrict_to_same_domain` is not enforced on nested `<sitemap>` entries

  intermediate · unassigned · bug · langchain · external

• #38719 Structured output with a raw JSON-schema dict is never validated, making ToolStrategy's handle_errors retry unreachable for dict schemas

  intermediate · unassigned · bug · langchain · external

• #38705 fix create_agent fixed settings when invoke dashscope models with customed response format

  intermediate · unassigned · bug · langchain · external

• #36128 Add support for Anthropic via Bedrock's `tool_search_tool_regex` in `ChatAnthropicBedrock`

  intermediate · unassigned · langchain · feature request · anthropic · external

• #34029 Tool calls sometimes fail due to missing positional argument

  intermediate · unassigned · documentation · bug · langchain · openai

• #38034 UserWarning: Pydantic serializer warnings when streaming with structured output

  intermediate · unassigned · bug · core · langchain · openai

• #32197 Support structured output with Pydantic models in `langchain-huggingface`

  intermediate · unassigned · help wanted · investigate · integration · huggingface

• #38718 PIIMiddleware: PIIMatch not exported from public package, and docs use wrong field names for custom detectors

  intermediate · unassigned · bug · langchain · external

• #38779 ChatAnthropic.bind_tools mutates caller-provided tool_choice dictionary

  intermediate · unassigned · bug · langchain · anthropic · external

• #34328 OpenRouter / LiteLLM / OpenAI API compatibility across providers / proxies

  intermediate · assigned · internal · openrouter · vllm

• #38768 feat(middleware): AskQuestionMiddleware for structured HITL questions

  intermediate · unassigned · langchain · feature request · external

• #36853 Architecture Proposal: Hardening PythonREPLTool Local Execution via Defense-in-Depth (AST + PEP 578)

  intermediate · unassigned · feature request · external

• #38713 `convert_runnable_to_tool` wraps TypedDict input in `RootModel` causing OpenAI tool calls to fail

  intermediate · unassigned · bug · core · langchain · external

• #38764 [FEATURE] Preserve reasoning / reasoning_content field in streaming ChatOpenAI chunks

  intermediate · unassigned · feature request · openai · external

• #38639 ChatAnthropic streaming drops thinking field from empty thinking blocks (signature_delta-only), replay fails with 400 thinking.thinking: Field required

  intermediate · assigned · bug · anthropic · external

Showing 30 of 100 open issues — the rest are on GitHub.