Releases·Tool Release
LangChain September 2026 release adds delegated LangSmith access, stop_reason SDK tracking, security patches
LangChain's September 2026 release updated its Python and JavaScript packages with delegated LangSmith access for sandboxes, SDK run tracking that captures stop_reason values, and security patches including a soupsieve ReDoS fix. Companion packages LangChain OpenAI (v1.6.5), LangChain Anthropic (v1.7.4), and LangGraph CLI (v0.4.32) also shipped updates. The changes improve observability and security for developers running agentic workflows in production.
4 engagement1 sourceThu, Sep 24, 2026, 03:15 AM
AI Overview
- September 2026 release adds delegated LangSmith access for sandboxes, authenticating via LangSmith service URLs with JavaScript integration support.
- SDK run tracking now captures stop_reason values for Claude SDK runs, plus tool/response ID logging and Python agent-scoped addressing.
- Security patches fix soupsieve ReDoS vulnerabilities, reducing denial-of-service exposure across Python and JavaScript dependency updates.
- LangChain OpenAI (v1.6.5) and Anthropic (v1.7.4) add mid-conversation tool changes and model profiles for Opus 5.5 and GPT-6.
- LangGraph CLI (v0.4.32) introduces self-hosted deployment improvements, image URI flags, and clarified agent configuration defaults.
Related
Releases·Tool Release
LangChain 1.3.15 adds trace_policy to AgentMiddleware
LangChain released version 1.3.15 with new features and fixes, including exposing trace_policy on AgentMiddleware and preserving history on SummarizationMiddleware summary failure. The release also includes type error fixes and dependency bumps.
LangChain released version 1.3.15 with new features and fixes, including exposing trace_policy on AgentMiddleware and preserving history on SummarizationMiddleware summary failure. The release also includes type error fixes and dependency bumps.
0 engagement1 sourcegithubTue, Aug 11, 2026, 07:11 PM
Social·Community
Intern solo-builds internal AI assistant for semiconductor org, sharing architecture lessons from RAG to MCP deployment
An intern at a semiconductor company describes building an internal AI assistant alone, with no senior AI engineer to review decisions. The system pairs a source-aware deterministic RAG knowledge assistant with a trained classifier for ticket routing, plus custom observability built in-house due to data-residency constraints. Stage-level latency logging cut p99 latency from ~70 seconds to ~25.
An intern at a semiconductor company describes building an internal AI assistant alone, with no senior AI engineer to review decisions. The system pairs a source-aware deterministic RAG knowledge assistant with a trained classifier for ticket routing, plus custom observability built in-house due to data-residency constraints. Stage-level latency logging cut p99 latency from ~70 seconds to ~25.
33 engagement1 sourceredditSun, Sep 20, 2026, 01:36 AM
Social·Community
Users seek LangGraph observability alternatives to LangSmith
A developer asks for observability tools compatible with LangChain and LangGraph, citing LangSmith as too expensive and self-hosted Langfuse as too burdensome for hobby projects. Suggestions include Brainstrust, Arize, Logfire, and Honeycomb.
A developer asks for observability tools compatible with LangChain and LangGraph, citing LangSmith as too expensive and self-hosted Langfuse as too burdensome for hobby projects. Suggestions include Brainstrust, Arize, Logfire, and Honeycomb.
12 engagement1 sourceredditThu, Jul 30, 2026, 11:52 AM
Releases·Tool Release
LangChain releases langchain-anthropic 1.7.0 with skills container support and 1.0 SDK
LangChain shipped langchain-anthropic 1.7.0, adding top-level skill parameters via `container`, a new thinking display mode, and support for the Anthropic 1.0 SDK. The release also fixes beta-header handling for the advisor tool and surfaces gateway response metadata. Practitioners using LangChain with Claude models gain updated integration features and fixes.
LangChain shipped langchain-anthropic 1.7.0, adding top-level skill parameters via `container`, a new thinking display mode, and support for the Anthropic 1.0 SDK. The release also fixes beta-header handling for the advisor tool and surfaces gateway response metadata. Practitioners using LangChain with Claude models gain updated integration features and fixes.
0 engagement1 sourcegithubThu, Aug 27, 2026, 03:24 PM
Releases·Tool Release
LangChain releases v1.3.12 with interrupt propagation and shell middleware fix
LangChain released version 1.3.12, fixing interrupt propagation through ToolRetryMiddleware and avoiding shared process-group kill in shell middleware. The companion langchain-openai v1.3.4 suppresses Pydantic serializer warnings on structured output parsed fields.
LangChain released version 1.3.12, fixing interrupt propagation through ToolRetryMiddleware and avoiding shared process-group kill in shell middleware. The companion langchain-openai v1.3.4 suppresses Pydantic serializer warnings on structured output parsed fields.
4 engagement2 sourcesgithubWed, Jul 8, 2026, 10:38 PM