# LangGraph

**URL:** https://forum.langchain.com/c/oss-product-help-lc-and-lg/langgraph/13.md

[Latest](https://forum.langchain.com/latest.md) · [Categories](https://forum.langchain.com/categories.md) · [Tags](https://forum.langchain.com/tags.md)

---

## [About the LangGraph category](https://forum.langchain.com/t/about-the-langgraph-category/37)

<div class="topic-metadata">

**Author:** [@jess](https://forum.langchain.com/u/jess)\
**Replies:** 1\
**Last updated:** [May 29, 2026, 3:36pm UTC](https://forum.langchain.com/t/about-the-langgraph-category/37 "2026-05-29T15:36:38Z")

</div>

Questions, issues, and/or discussions regarding LangGraph.

---

## [Should I choose typescript or python for this project , as I have experience in JS](https://forum.langchain.com/t/should-i-choose-typescript-or-python-for-this-project-as-i-have-experience-in-js/5255)

<div class="topic-metadata">

**Author:** [@Prad\_S](https://forum.langchain.com/u/Prad_S)\
**Replies:** 4\
**Last updated:** [November 15, 2026, 8:09pm UTC](https://forum.langchain.com/t/should-i-choose-typescript-or-python-for-this-project-as-i-have-experience-in-js/5255 "2026-11-15T20:09:48Z")

</div>

I’m building a production-level multi-agent system that automates software development workflows — from understanding development tickets and analyzing codebases to writing code inside isolated Docker sandboxes, running …

---

## [A checkpointer that accepts two concurrent writers still scores conformance\_level = FULL](https://forum.langchain.com/t/a-checkpointer-that-accepts-two-concurrent-writers-still-scores-conformance-level-full/4460)

<div class="topic-metadata">

**Author:** [@bharatnpti](https://forum.langchain.com/u/bharatnpti)\
**Replies:** 17\
**Last updated:** [November 15, 2026, 4:33pm UTC](https://forum.langchain.com/t/a-checkpointer-that-accepts-two-concurrent-writers-still-scores-conformance-level-full/4460 "2026-11-15T16:33:30Z")

</div>

I ran the checkpoint-conformance suite from main against three savers: InMemorySaver AsyncPostgresSaver 3.1.2 one I wrote that deliberately accepts writes from any number of workers on the same thread All three pass. …

---

## [How are you guys stopping agents from silent loops/stampedes? Standard step\_counters feel like garbage](https://forum.langchain.com/t/how-are-you-guys-stopping-agents-from-silent-loops-stampedes-standard-step-counters-feel-like-garbage/4479)

<div class="topic-metadata">

**Author:** [@minakshihub](https://forum.langchain.com/u/minakshihub)\
**Replies:** 7\
**Last updated:** [November 14, 2026, 6:02pm UTC](https://forum.langchain.com/t/how-are-you-guys-stopping-agents-from-silent-loops-stampedes-standard-step-counters-feel-like-garbage/4479 "2026-11-14T18:02:26Z")

</div>

I’ve been looking at multi-agent failure states, and it seems like everyone is relying on basic Python step\_counters or LLM prompt-engineering (“do not repeat yourself”) to prevent agents from falling into infinite tool-…

---

## [30 open LangGraph and LangGraph.js bugs in interrupts, retries, checkpoints and caching, most with a community fix waiting](https://forum.langchain.com/t/30-open-langgraph-and-langgraph-js-bugs-in-interrupts-retries-checkpoints-and-caching-most-with-a-community-fix-waiting/5135)

<div class="topic-metadata">

**Author:** [@yifanxiong272](https://forum.langchain.com/u/yifanxiong272)\
**Replies:** 3\
**Last updated:** [November 14, 2026, 8:47am UTC](https://forum.langchain.com/t/30-open-langgraph-and-langgraph-js-bugs-in-interrupts-retries-checkpoints-and-caching-most-with-a-community-fix-waiting/5135 "2026-11-14T08:47:08Z")

</div>

Hi LangGraph team, Over the past weeks we reported 30 bugs in the durable-execution features that applications rely on to pause for a human, survive failures and resume: interrupts and resumes, retry and error-handler p…

---

## [How to inject ToolRuntime if I call a tool manually?](https://forum.langchain.com/t/how-to-inject-toolruntime-if-i-call-a-tool-manually/5233)

<div class="topic-metadata">

**Author:** [@MSJavaScript](https://forum.langchain.com/u/MSJavaScript)\
**Replies:** 3\
**Last updated:** [November 13, 2026, 3:35pm UTC](https://forum.langchain.com/t/how-to-inject-toolruntime-if-i-call-a-tool-manually/5233 "2026-11-13T15:35:39Z")

</div>

Hello, Here is such a node in my graph: This node handle all the interrupts, but some tools have ToolRuntime in their args, how can I inject this into the tool call ? Thanks. def create\_interrupt\_node(model:BaseChatMod…

---

## [Best way to handle very large tool outputs in LangGraph (avoid LLM + LangSmith overload)](https://forum.langchain.com/t/best-way-to-handle-very-large-tool-outputs-in-langgraph-avoid-llm-langsmith-overload/2445)

<div class="topic-metadata">

**Author:** [@csarww](https://forum.langchain.com/u/csarww)\
**Replies:** 1\
**Last updated:** [November 13, 2026, 2:10pm UTC](https://forum.langchain.com/t/best-way-to-handle-very-large-tool-outputs-in-langgraph-avoid-llm-langsmith-overload/2445 "2026-11-13T14:10:27Z")

</div>

Hi all — I’m building an agentic workflow (LangGraph + Gemini) where one of my tools returns a very large JSON output. The full data is required for downstream processing, but: I must not send the raw JSON to the LLM …

---

## [Feature proposal: A reusable reducer for deduplicating parallel results in LangGraph Python](https://forum.langchain.com/t/feature-proposal-a-reusable-reducer-for-deduplicating-parallel-results-in-langgraph-python/5144)

<div class="topic-metadata">

**Author:** [@harrylettering](https://forum.langchain.com/u/harrylettering)\
**Replies:** 1\
**Last updated:** [November 12, 2026, 5:24pm UTC](https://forum.langchain.com/t/feature-proposal-a-reusable-reducer-for-deduplicating-parallel-results-in-langgraph-python/5144 "2026-11-12T17:24:31Z")

</div>

In a LangGraph workflow that fans out to multiple retrievers, two branches can return the same document ID. A custom Annotated reducer can merge those updates, but the same deduplication logic has to be written for each …

---

## [Human-in-the-loop approval dashboard for LangGraph agents — open source, free to deploy](https://forum.langchain.com/t/human-in-the-loop-approval-dashboard-for-langgraph-agents-open-source-free-to-deploy/3616)

<div class="topic-metadata">

**Author:** [@surya\_02](https://forum.langchain.com/u/surya_02)\
**Replies:** 5\
**Last updated:** [November 8, 2026, 4:33am UTC](https://forum.langchain.com/t/human-in-the-loop-approval-dashboard-for-langgraph-agents-open-source-free-to-deploy/3616 "2026-11-08T04:33:12Z")

</div>

I’ve been building LangGraph agents and kept hitting the same problem: no clean way to pause an agent before a sensitive action and wait for a human to approve or reject it before it executes. LangGraph has interrupt() …

---

## [\`interrupt()\` records that a human resumed, not which human](https://forum.langchain.com/t/interrupt-records-that-a-human-resumed-not-which-human/4966)

<div class="topic-metadata">

**Author:** [@TroyClifford](https://forum.langchain.com/u/TroyClifford)\
**Replies:** 10\
**Last updated:** [November 7, 2026, 3:36pm UTC](https://forum.langchain.com/t/interrupt-records-that-a-human-resumed-not-which-human/4966 "2026-11-07T15:36:05Z")

</div>

Something I ran into building an approval flow, and the workaround I settled on. My actual question is at the bottom: I want to know whether there is an idiom for this that I have missed. interrupt() pauses the graph …

---

## [Optimizing RAG Retrieval Latency and Token Overhead in Production LangChain Pipelines](https://forum.langchain.com/t/optimizing-rag-retrieval-latency-and-token-overhead-in-production-langchain-pipelines/5047)

<div class="topic-metadata">

**Author:** [@softuvo](https://forum.langchain.com/u/softuvo)\
**Replies:** 3\
**Last updated:** [November 7, 2026, 7:09am UTC](https://forum.langchain.com/t/optimizing-rag-retrieval-latency-and-token-overhead-in-production-langchain-pipelines/5047 "2026-11-07T07:09:56Z")

</div>

Hi LangChain Community, While architecting enterprise RAG applications at Softuvo, our engineering team ran into latency bottlenecks and high token costs when retrieving and re-ranking complex multi-document contexts. T…

---

## [Are people hitting race conditions in multi-agent LangChain setups?](https://forum.langchain.com/t/are-people-hitting-race-conditions-in-multi-agent-langchain-setups/3202)

<div class="topic-metadata">

**Author:** [@Thorwig](https://forum.langchain.com/u/Thorwig)\
**Replies:** 7\
**Last updated:** [November 6, 2026, 12:19pm UTC](https://forum.langchain.com/t/are-people-hitting-race-conditions-in-multi-agent-langchain-setups/3202 "2026-11-06T12:19:54Z")

</div>

I’ve been spending time on multi-agent workflows lately, and I keep coming back to the same question: once multiple agents or tool calls can mutate the same file / DB row / shared state, what is the expected safety mode…

---

## [Enhancement: Allow configuring Redis host/URL via environment variables in test\_redis\_cache.py](https://forum.langchain.com/t/enhancement-allow-configuring-redis-host-url-via-environment-variables-in-test-redis-cache-py/5029)

<div class="topic-metadata">

**Author:** [@wuzm2k05](https://forum.langchain.com/u/wuzm2k05)\
**Replies:** 1\
**Last updated:** [November 4, 2026, 2:13pm UTC](https://forum.langchain.com/t/enhancement-allow-configuring-redis-host-url-via-environment-variables-in-test-redis-cache-py/5029 "2026-11-04T14:13:16Z")

</div>

Summary In libs/checkpoint/tests/test\_redis\_cache.py, the setup fixture currently hardcodes host="localhost" when initializing the redis.Redis client. In my development environment, I have a dedicated Redis instance per…

---

## [Review request: is this a fair LangGraph setup for a "blog post to social posts" benchmark?](https://forum.langchain.com/t/review-request-is-this-a-fair-langgraph-setup-for-a-blog-post-to-social-posts-benchmark/4847)

<div class="topic-metadata">

**Author:** [@kelvindesman](https://forum.langchain.com/u/kelvindesman)\
**Replies:** 1\
**Last updated:** [November 3, 2026, 8:08pm UTC](https://forum.langchain.com/t/review-request-is-this-a-fair-langgraph-setup-for-a-blog-post-to-social-posts-benchmark/4847 "2026-11-03T20:08:39Z")

</div>

Hi all. We’re comparing four ways to turn one blog post into a LinkedIn post, an X thread and a 5-slide carousel: a plain cron job, cron plus one LLM call, a LangGraph graph, and an autonomous agent (Hermes Agent). Same …

---

## [Creating AI model that remembers and more importantly forgets and compartmentalises information closer to that of AGI and how a humans brain works](https://forum.langchain.com/t/creating-ai-model-that-remembers-and-more-importantly-forgets-and-compartmentalises-information-closer-to-that-of-agi-and-how-a-humans-brain-works/4366)

<div class="topic-metadata">

**Author:** [@RobertORourke](https://forum.langchain.com/u/RobertORourke)\
**Replies:** 2\
**Last updated:** [November 4, 2026, 8:25pm UTC](https://forum.langchain.com/t/creating-ai-model-that-remembers-and-more-importantly-forgets-and-compartmentalises-information-closer-to-that-of-agi-and-how-a-humans-brain-works/4366 "2026-11-04T20:25:15Z")

</div>

Architectural Proposal: Dynamic Topic-Segmented Memory (DTSM) for Large Language Models Executive Summary Current Large Language Model (LLM) architectures struggle with long-form, multi-topic conversations. Standard co…

---

## [Proposal: checkpoint aware context management for long-running agents (+ optional KV-cache editing)](https://forum.langchain.com/t/proposal-checkpoint-aware-context-management-for-long-running-agents-optional-kv-cache-editing/4792)

<div class="topic-metadata">

**Author:** [@Anipaleja](https://forum.langchain.com/u/Anipaleja)\
**Replies:** 4\
**Last updated:** [October 29, 2026, 1:46am UTC](https://forum.langchain.com/t/proposal-checkpoint-aware-context-management-for-long-running-agents-optional-kv-cache-editing/4792 "2026-10-29T01:46:03Z")

</div>

Hi maintainers, Ive been working on a context management layer for LangGraph and wanted to get some feedback before opening a PR. LangGraph already has tools like pre\_model\_hook, RemoveMessage, and trim\_messages, but I …

---

## [Production use of PostgresSaver: service-owned migrations and checkpoint retention in Node.js](https://forum.langchain.com/t/production-use-of-postgressaver-service-owned-migrations-and-checkpoint-retention-in-node-js/4481)

<div class="topic-metadata">

**Author:** [@waiz-ksolves](https://forum.langchain.com/u/waiz-ksolves)\
**Replies:** 3\
**Last updated:** [October 28, 2026, 6:38pm UTC](https://forum.langchain.com/t/production-use-of-postgressaver-service-owned-migrations-and-checkpoint-retention-in-node-js/4481 "2026-10-28T18:38:50Z")

</div>

We are planning to use @langchain/langgraph-checkpoint-postgres in a production Node.js service and have a few questions about the recommended database ownership and retention model. Our service team normally owns all P…

---

## [Trying to understand how to optimise on agent cost](https://forum.langchain.com/t/trying-to-understand-how-to-optimise-on-agent-cost/4909)

<div class="topic-metadata">

**Author:** [@raghusiddarth](https://forum.langchain.com/u/raghusiddarth)\
**Replies:** 3\
**Last updated:** [October 28, 2026, 6:15pm UTC](https://forum.langchain.com/t/trying-to-understand-how-to-optimise-on-agent-cost/4909 "2026-10-28T18:15:47Z")

</div>

Hi, We’ve built an agent with langchain, I’m spending around $0.9 per session - How do I optimse on this? any patterns that I can look for?

---

## [Proposal: provider-neutral typed decisions for LangGraph Python and JS](https://forum.langchain.com/t/proposal-provider-neutral-typed-decisions-for-langgraph-python-and-js/4936)

<div class="topic-metadata">

**Author:** [@sjungwon03](https://forum.langchain.com/u/sjungwon03)\
**Replies:** 0\
**Last updated:** [October 28, 2026, 5:04pm UTC](https://forum.langchain.com/t/proposal-provider-neutral-typed-decisions-for-langgraph-python-and-js/4936 "2026-10-28T17:04:01Z")

</div>

I’d like to connect dedicated decision models to existing LangGraph LLM workflows in both Python and JS. A decision should be able to request any subset of named Choice, Noul, and Score judgments, then use the result eit…

---

## [How should enterprise AI agents cache, refresh, and select skills in runtime? ](https://forum.langchain.com/t/how-should-enterprise-ai-agents-cache-refresh-and-select-skills-in-runtime/4845)

<div class="topic-metadata">

**Author:** [@suresandeep03](https://forum.langchain.com/u/suresandeep03)\
**Replies:** 2\
**Last updated:** [October 28, 2026, 3:43am UTC](https://forum.langchain.com/t/how-should-enterprise-ai-agents-cache-refresh-and-select-skills-in-runtime/4845 "2026-10-28T03:43:41Z")

</div>

Hi there, I’m are building an AI Agent that uses skill.md files. Each skill.md contains: Skill metadata skill\_name description examples business domain tags Execution instructions SQL query itself Snowflake execut…

---

## [Feature Proposal: First-Class AgentRouter and LLMRouter Abstractions](https://forum.langchain.com/t/feature-proposal-first-class-agentrouter-and-llmrouter-abstractions/4915)

<div class="topic-metadata">

**Author:** [@yogeshkulkarni11](https://forum.langchain.com/u/yogeshkulkarni11)\
**Replies:** 0\
**Last updated:** [October 27, 2026, 5:00pm UTC](https://forum.langchain.com/t/feature-proposal-first-class-agentrouter-and-llmrouter-abstractions/4915 "2026-10-27T17:00:34Z")

</div>

Feature Proposal: First-Class AgentRouter and LLMRouter Abstractions Background LangGraph provides excellent low-level primitives for building agent workflows, including nodes, edges, conditional routing, Command, state …

---

## [Proposal: governed collective-decision reference application for LangGraph](https://forum.langchain.com/t/proposal-governed-collective-decision-reference-application-for-langgraph/4907)

<div class="topic-metadata">

**Author:** [@Buffalo2024](https://forum.langchain.com/u/Buffalo2024)\
**Replies:** 0\
**Last updated:** [October 27, 2026, 4:36pm UTC](https://forum.langchain.com/t/proposal-governed-collective-decision-reference-application-for-langgraph/4907 "2026-10-27T16:36:06Z")

</div>

Problem LangGraph provides graph execution, state, routing, persistence, and human-in-the-loop primitives. A multi-agent decision graph can still accept malformed carriers, merge incompatible candidates, allow a reviewer…

---

## [Interrupt Inconsistency (LangGraph 1.x)](https://forum.langchain.com/t/interrupt-inconsistency-langgraph-1-x/1996)

<div class="topic-metadata">

**Author:** [@akbindal](https://forum.langchain.com/u/akbindal)\
**Replies:** 1\
**Last updated:** [October 27, 2026, 4:28pm UTC](https://forum.langchain.com/t/interrupt-inconsistency-langgraph-1-x/1996 "2026-10-27T16:28:51Z")

</div>

Seeing a mismatch in langgraph interrupt behavior after migrating to LangGraph 1.x. On local dev, interrupt shows up in thread\_state but not in the streaming. On Docker (OSS self-hosted), stream emits the interrupt but…

---

## [Create\_react\_agent occasionally not calling tools](https://forum.langchain.com/t/create-react-agent-occasionally-not-calling-tools/1873)

<div class="topic-metadata">

**Author:** [@FLFortunato](https://forum.langchain.com/u/FLFortunato)\
**Replies:** 2\
**Last updated:** [October 27, 2026, 3:19pm UTC](https://forum.langchain.com/t/create-react-agent-occasionally-not-calling-tools/1873 "2026-10-27T15:19:55Z")

</div>

Hey there, everyone! I’m facing a problem with my agent built with create\_react\_agent. I start a conversation questioning about some infos and it calls the tools correctly, but sometimes it just stops invoking the tools…

---

## [Proposal: additional docs for implementing custom DB checkpointers or a guide on generic base checkpointer](https://forum.langchain.com/t/proposal-additional-docs-for-implementing-custom-db-checkpointers-or-a-guide-on-generic-base-checkpointer/3735)

<div class="topic-metadata">

**Author:** [@Wonder-donbury](https://forum.langchain.com/u/Wonder-donbury)\
**Replies:** 3\
**Last updated:** [October 23, 2026, 6:58pm UTC](https://forum.langchain.com/t/proposal-additional-docs-for-implementing-custom-db-checkpointers-or-a-guide-on-generic-base-checkpointer/3735 "2026-10-23T18:58:26Z")

</div>

Currently I’m guessing a lot of agent PoC made with LangGraph is preparing to get to production level, and so is my project, and some projects that needs to be deployed on-prem, database licensing of the client’s side is…

---

## [How to pass a "runtime args" to a graph node?](https://forum.langchain.com/t/how-to-pass-a-runtime-args-to-a-graph-node/4750)

<div class="topic-metadata">

**Author:** [@MSJavaScript](https://forum.langchain.com/u/MSJavaScript)\
**Replies:** 2\
**Last updated:** [October 15, 2026, 7:42am UTC](https://forum.langchain.com/t/how-to-pass-a-runtime-args-to-a-graph-node/4750 "2026-10-15T07:42:36Z")

</div>

Hello. I have a planner node, and I want it accept some runtime args, for example: def planner\_node(state:MainState, args:dict): pass This node may make (1) A fresh plan, without any dependence of previous steps. …

---

## [Extending checkpoint schema with user\_id/agent\_id for cross-thread querying — trade-offs vs Thread Registry?](https://forum.langchain.com/t/extending-checkpoint-schema-with-user-id-agent-id-for-cross-thread-querying-trade-offs-vs-thread-registry/4429)

<div class="topic-metadata">

**Author:** [@santosh-nallur](https://forum.langchain.com/u/santosh-nallur)\
**Replies:** 6\
**Last updated:** [October 7, 2026, 2:28pm UTC](https://forum.langchain.com/t/extending-checkpoint-schema-with-user-id-agent-id-for-cross-thread-querying-trade-offs-vs-thread-registry/4429 "2026-10-07T14:28:50Z")

</div>

Hi Team, I want to gather opinion on one of our requirements related to LangGraph. I found a similar discussion here where this is discussed at length and provides useful insights, but I have more specific questions. U…

---

## [What is the right way to conditionally interrupt](https://forum.langchain.com/t/what-is-the-right-way-to-conditionally-interrupt/4637)

<div class="topic-metadata">

**Author:** [@MSJavaScript](https://forum.langchain.com/u/MSJavaScript)\
**Replies:** 4\
**Last updated:** [October 1, 2026, 4:26pm UTC](https://forum.langchain.com/t/what-is-the-right-way-to-conditionally-interrupt/4637 "2026-10-01T16:26:22Z")

</div>

The langgraph manual said " Do not conditionally skip interrupt calls within a node", but I do have some situations in which I need to conditionally interrupt. For example, I have a very complex tool def generate\_elemen…

---

## [How to handle growing checkpoint blob?](https://forum.langchain.com/t/how-to-handle-growing-checkpoint-blob/4448)

<div class="topic-metadata">

**Author:** [@hks](https://forum.langchain.com/u/hks)\
**Replies:** 3\
**Last updated:** [September 23, 2026, 2:08pm UTC](https://forum.langchain.com/t/how-to-handle-growing-checkpoint-blob/4448 "2026-09-23T14:08:46Z")

</div>

I’m using open-source LangGraph with PostgresSaver and durability: "exit". The checkpoints table looks fine for my use case since I’m seeing roughly one checkpoint per chat request. The issue is that checkpoint\_blobs ke…

---

## [Question on Open Github Issue #7345: Status of configurable PostgreSQL schema](https://forum.langchain.com/t/question-on-open-github-issue-7345-status-of-configurable-postgresql-schema/4558)

<div class="topic-metadata">

**Author:** [@alive](https://forum.langchain.com/u/alive)\
**Replies:** 1\
**Last updated:** [September 18, 2026, 2:28pm UTC](https://forum.langchain.com/t/question-on-open-github-issue-7345-status-of-configurable-postgresql-schema/4558 "2026-09-18T14:28:22Z")

</div>

Hey everyone! :waving\_hand: Quick question regarding Issue #7345 (“Configurable PostgreSQL schema for langgraph-checkpoint-postgres”): https://github.com/langchain-ai/langgraph/issues/7345 It looks like there has been…

[Next page](https://forum.langchain.com/c/oss-product-help-lc-and-lg/langgraph/13.md?page=1)
