fix(langgraph): carry ToolMessage.error onto the LangChain status flag - #2263
Merged
contextablemark merged 1 commit intoAug 3, 2026
Conversation
Both langgraph adapters dropped AG-UI's ToolMessage.error when converting an incoming tool message to a LangChain ToolMessage. LangChain has a status field (success/error) for this, and providers such as langchain_anthropic map it onto the model's tool-result flag. With the field dropped, a client-reported tool failure reached the model as a success. The impact is provider-dependent, so it was invisible on Gemini. Map error onto status in both adapters (TypeScript and Python): status is error when the AG-UI error field is set, else success. A test is added on each side. Scope: the status flag only. Preserving the error text and round-tripping it back to the client are separate questions raised in the issue and left out here. Fixes ag-ui-protocol#2226
This was referenced Aug 3, 2026
Contributor
Python Preview PackagesVersion
Install with uvAdd the TestPyPI index to your [[tool.uv.index]]
name = "testpypi"
url = "https://test.pypi.org/simple/"
explicit = trueThen install the packages you need: # Core SDK
uv add 'ag-ui-protocol==0.0.0.dev1785306123' --index testpypi
# Integrations (each already depends on the matching ag-ui-protocol preview)
uv add 'ag-ui-langgraph==0.0.0.dev1785306123' --index testpypi
uv add 'ag-ui-crewai==0.0.0.dev1785306123' --index testpypi
# NOTE: ag-ui-agent-spec depends on pyagentspec (git-only, not on PyPI).
# You will need to install pyagentspec separately from its git repo.
uv add 'ag-ui-agent-spec==0.0.0.dev1785306123' --index testpypi
uv add 'ag_ui_adk==0.0.0.dev1785306123' --index testpypi
uv add 'ag_ui_strands==0.0.0.dev1785306123' --index testpypiInstall with pippip install \
--index-url https://test.pypi.org/simple/ \
--extra-index-url https://pypi.org/simple/ \
ag-ui-protocol==0.0.0.dev1785306123
Commit: e8a4ba0 |
Contributor
|
Reviewed and verified locally against the PR head.
The fix is correct, minimal, and tested on both sides. I've filed the two carve-outs so the scope of this PR is recorded and the remaining work isn't lost:
On the "should the |
@ag-ui/a2a-middleware
@ag-ui/a2ui-middleware
@ag-ui/event-throttle-middleware
@ag-ui/mcp-apps-middleware
@ag-ui/mcp-middleware
@ag-ui/a2a
@ag-ui/adk
@ag-ui/ag2
@ag-ui/agno
@ag-ui/aws-strands
@ag-ui/crewai
@ag-ui/langchain
@ag-ui/claude-agent-sdk
@ag-ui/langgraph
@ag-ui/llamaindex
@ag-ui/mastra
@ag-ui/pydantic-ai
@ag-ui/vercel-ai-sdk
@ag-ui/watsonx
@ag-ui/a2ui-toolkit
create-ag-ui-app
@ag-ui/client
@ag-ui/core
@ag-ui/encoder
@ag-ui/proto
commit: |
contextablemark
approved these changes
Aug 3, 2026
contextablemark
left a comment
contextablemark
left a comment
Contributor
There was a problem hiding this comment.
Looks good. Created some follow-on issues.
This was referenced Aug 5, 2026
BenTaylorDev
pushed a commit
that referenced
this pull request
Aug 10, 2026
….error The reverse conversion (langchainMessagesToAgui / langchain_messages_to_agui) dropped the LangChain tool result status, so a status='error' tool message came back to the client with error unset. After #2263 fixed the forward direction, a client whose message list seeds the next run (fresh thread, stateless replay, snapshot-then-resend) re-converted with error unset and silently produced status='success' again, undoing the fix. Map status == 'error' back onto AG-UI's error in both adapters. The forward direction is flag-only, so the original text is not recoverable here; error carries a fixed sentinel so a client can tell a reported failure apart from a reported failure text. Flag durable, text best-effort, per the standard agreed on #2306. A test is added on each side. Fixes #2305
BenTaylorDev
pushed a commit
that referenced
this pull request
Aug 10, 2026
…in four more adapters The same one-line omission fixed for LangGraph in #2263 is present in several other first-party adapters: AG-UI's ToolMessage.error is not read when an incoming tool result is converted into the target framework's tool-result shape, so a client-reported tool failure is handed to the model as a success. Per the standard agreed on #2306, set the framework's error flag from bool(error) (flag only; text is not folded because every target here has a dedicated flag): langchain/typescript ToolMessage status (verbatim #2263); aws-strands/python Bedrock toolResult status (was pinned to success); vercel-ai-sdk/typescript AI SDK v4 tool-result isError; mastra/typescript AI SDK v4 tool-result isError. Both vercel-ai-sdk and mastra resolve AI SDK v4, whose ToolResultPart already carries isError; no version gate is needed. Also fix aws-strands _build_snapshot_messages, an AG-UI to AG-UI rebuild of the client's own messages: it dropped the client's error and encrypted_value on the snapshot echo. Copy both through. A test is added next to each adapter's conversion tests. Fixes #2306
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #2226.
Problem
Both langgraph adapters drop AG-UI's
ToolMessage.errorwhen they convert an incoming tool message to a LangChainToolMessage. LangChain has astatusfield ("success" | "error") for exactly this, and providers such aslangchain_anthropicmap it onto the model's tool-result flag. With the field dropped, a client-reported tool failure reaches the model as a success. The impact is provider-dependent, so it is invisible on Gemini and shows up on Anthropic.Fix
Map
errorontostatusin both adapters:"error"when the AG-UIerrorfield is set, else"success".integrations/langgraph/typescript/src/utils.ts, thetoolbranch ofaguiMessagesToLangChain. The langgraph-sdkToolMessagetype already typesstatus?: "error" | "success".integrations/langgraph/python/ag_ui_langgraph/utils.py, thetoolbranch ofagui_messages_to_langchain.The AG-UI
ToolMessageschema already haserrorin both languages, so no type changes are needed.Scope
This PR maps the flag only. The two open questions from the issue — preserving the
errortext and round-tripping it back to the client — look like a separate protocol topic and are left out here.Tests
A test is added on each side, next to the existing tool-message conversion tests.
message-conversion.test.ts— full package suite green, 265 tests.test_message_conversion.py—unittest discover testsgreen, 43 tests.