langchain-ai
langchain
Blog
Docs
Changelog
Blog
Docs
Changelog
Overview
Branches
Benchmarks
Runs
Performance History
Latest Results
chore(langchain,langchain-classic): uncomment optional deps (#37163) Re-enable the `[community]`, `[azure-ai]`, and `[cohere]` extras on `langchain-classic`, and the `[cohere]` extra on `langchain` (v1). These had been commented out as a temporary workaround during the `langchain` -> `langchain-classic` rename so the renamed package could ship before downstream partners were re-released against it. Now that `langchain-community` 0.4.1, `langchain-cohere` 0.5.1, and `langchain-azure-ai` 1.2.3 are published with the correct dependency targets, the extras can be restored.
master
53 minutes ago
chore(langchain,langchain-classic): uncomment optional deps
mdrxy/fix-pyprojects
1 hour ago
feat(openai): support apply_patch built-in tool
nidhi1603:feat/openai-apply-patch-tool
12 hours ago
feat(openai): support apply_patch built-in tool
nidhi1603:feat/openai-apply-patch-tool
14 hours ago
release(langchain): 1.3.0a1
nh/release-langchain-1.3.0a1
2 days ago
bump langgraph floor to 1.2.0a5 to pull in prebuilt 1.1.0a2 scope-filter fix
nh/streaming-langchain-v1.4
2 days ago
address review: tighten transformers= type, rename V2 -> V3, assert ordering, add nested-agent test - Mirror langgraph's `Sequence[Callable[[tuple[str, ...]], Any]]` type for the `transformers=` kwarg on `create_agent` so the scope-aware factory contract surfaces in IDE hover. - Rename test classes from `TestAgentStreamV2*` to `TestAgentStreamV3*`; the file has tested v3 since the upstream API rename. - In `test_caller_transformers_appended_not_replaced`, assert that `ToolCallTransformer` is registered before any user-supplied transformer in the mux pipeline — pinning the "appended after" contract. - Add `test_inner_agent_tool_calls_via_subgraph` covering the canonical nested-agent pattern: a `create_agent` invoked through a `@tool` shows up as a subgraph on the outer run, and its own tool calls are reachable via `run.subgraphs[*].tool_calls`. This requires the per-scope `ToolCallTransformer` namespace filter in https://github.com/langchain-ai/langgraph/pull/7681; it fails against langgraph-prebuilt 1.1.0a1 and passes against 1.1.0a2+.
nh/streaming-langchain-v1.4
2 days ago
feat(langchain): wire stream_events(version='v3') into create_agent Brings the v3 streaming protocol to `langchain.agents.create_agent` so callers can consume content-block events and typed projections through the new `Pregel.stream_events(version="v3")` / `astream_events(version="v3")` paths. - `create_agent` accepts a `transformers=` parameter that's appended after the built-in `ToolCallTransformer`, registered on the compiled graph so subgraph mini-muxes get fresh per-scope instances. - Bumps `langchain-core>=1.4.0a2,<2.0.0` (the v3 protocol shipped in 1.4.0a2 with the unwound `astream_events` return type). - Bumps `langgraph>=1.2.0a4,<1.3.0` (1.2.0a4 widened the `langgraph-prebuilt` constraint to pull in 1.1.0a1 which exposes `ToolCallTransformer`). - `[tool.uv] prerelease = "allow"` so transitive prerelease constraints (`langgraph-checkpoint>=4.1.0a3`) resolve under the same mode `uv lock` and `uv run` use. - Replaces the `nh/streaming-for-alpha-release` git pin and the `override-dependencies = ["langchain-core"]` workaround now that the upstream alphas are published. - New `tests/unit_tests/agents/test_agent_streaming.py` covering sync + async v3 dispatch through `create_agent`. PR 3 of 4 in the streaming alpha-release chain (langchain-ai/langchain#37111, langchain-ai/langgraph#7677, langchain-ai/langgraph#7679).
nh/streaming-langchain-v1.4
2 days ago
Latest Branches
CodSpeed Performance Gauge
0%
chore(langchain,langchain-classic): uncomment optional deps
#37163
1 hour ago
bf84c1d
mdrxy/fix-pyprojects
CodSpeed Performance Gauge
+1%
feat(openai): support apply_patch built-in tool
#37157
14 hours ago
3a34a22
nidhi1603:feat/openai-apply-patch-tool
CodSpeed Performance Gauge
0%
release(langchain): 1.3.0a1
#37140
2 days ago
dbae0c2
nh/release-langchain-1.3.0a1
© 2026 CodSpeed Technology
Home
Terms
Privacy
Docs