github/langchain
1
0
Fork 0
mirror of https://github.com/hwchase17/langchain.git synced 2026-07-12 19:31:24 +00:00
Code Issues Packages Projects Releases Wiki Activity
14,412 Commits 1,187 Branches 1,307 Tags
d74b0cacfa8fd145957ff09a3e511aed5e9bfad4
Commit Graph

7685 Commits

Author SHA1 Message Date
Ernesto Revilla
d74b0cacfa fix(core): backport path-traversal fix to v0.3 (CVE-2026-34070, GHSA-qh6h-p6c9-ff54) (#37233) 2026-05-07 10:58:34 -04:00
Eugene Yurtsev
56bcb093ff Revert "x" 
This reverts commit 763b30e8b2.
 2026-05-07 10:51:43 -04:00
Eugene Yurtsev
763b30e8b2 x 2026-05-07 10:49:22 -04:00
Nick Hollon
8242df0742 release(langchain): 0.3.29 (#37212) 2026-05-05 16:57:53 -04:00
Nick Hollon
6efe96bea6 fix(langchain): restrict deserialization in langchain.storage._lc_store (#37209) 2026-05-05 16:29:40 -04:00
Nick Hollon
bba04da32b release(core): 0.3.85 (#37205) 2026-05-05 16:20:00 -04:00
Nick Hollon
083b199cb1 fix(core, langchain): harden load() against untrusted manifests (#37201) 2026-05-05 20:12:03 +00:00
Eugene Yurtsev
2517da67dc release(core): 0.3.84 (#36616) 
release 0.3.84
 2026-04-08 15:10:33 -04:00
Eugene Yurtsev
6bab0ba3c1 fix(core): sanitize prompts more (#36613) 
add more sanitization to prompts
 2026-04-08 14:32:25 -04:00
Eugene Yurtsev
4fe869f1c3 fix(langchain): bump minimum langchain-core to 0.3.73 (#35608) 
The within-batch deduplication fix (langchain-core 0.3.73, #32273)
changed `num_skipped` counting in the indexing API. Tests in
`libs/langchain` were updated to expect the new behavior, but the
minimum version was left at 0.3.72, causing [pre-release checks to
fail](https://github.com/langchain-ai/langchain/actions/runs/22783515938/job/66097466123).

Created with [Deep Agents
CLI](https://docs.langchain.com/oss/python/deepagents/cli/overview).
 2026-03-06 22:38:05 +00:00
Eugene Yurtsev
8d9ecb18a1 release(langchain): 0.3.28 (#35607) 
Patch release for langchain 0.3.28.

Created with [Deep Agents
CLI](https://docs.langchain.com/oss/python/deepagents/cli/overview).
 2026-03-06 16:52:45 -05:00
Eugene Yurtsev
b0add1f989 fix(langchain): backport patch ReDoS vulnerability in MRKL and ReAct action regex (CVE-2024-58340) (#35603) 
Backport of #35598 to the v0.3 branch. Patches the ReDoS vulnerability
in MRKL and ReAct agent action regex patterns (CVE-2024-58340).

Created with [Deep Agents
CLI](https://docs.langchain.com/oss/python/deepagents/cli/overview).
 2026-03-06 16:22:23 -05:00
Mason Daugherty
c0e2f08f78 release(core): 0.3.83 (#34733) 2026-01-12 20:16:58 -05:00
Angus Jelinek
4dab5fafc0 feat(core,langchain,text-splitters): (v0.3) use uuid7 for run ids (#34732) 
Backports #34172

---------

Co-authored-by: William FH <13333726+hinthornw@users.noreply.github.com>
Co-authored-by: Sydney Runkle <54324534+sydney-runkle@users.noreply.github.com>
Co-authored-by: Mason Daugherty <github@mdrxy.com>
 2026-01-12 20:09:58 -05:00
Mason Daugherty
e6dde2b99c release(core): 0.3.82 (#34688) 2026-01-09 13:30:14 -05:00
Mason Daugherty
126a337082 fix(core): defer persisting traces for iterator inputs (#34687) 
Backport of #34416
 2026-01-09 13:17:27 -05:00
Mason Daugherty
13c13c4bac feat(core): add usage_metadata to metadata in LangChainTracer (#34686) 
Backport of #34414
 2026-01-09 13:08:51 -05:00
Mason Daugherty
6dc06da1bb release(core): 0.3.81 (#34459) 2025-12-22 18:42:31 -06:00
Mason Daugherty
d9ec4c5cc7 fix(core): serialization patch (#34458) 
Backport of https://github.com/langchain-ai/langchain/pull/34455
 2025-12-22 18:41:31 -06:00
Eugene Yurtsev
640d85c60f release(core): 0.3.80 (#34039) 
Release 0.3.80
 2025-11-19 16:53:03 -05:00
Eugene Yurtsev
fa7789d6c2 fix(core): fix validation for input variables in f-string templates, restrict functionality supported by jinja2, mustache templates (#34038) 
* Fix validation for input variables in f-string templates
* Restrict functionality of features supported by jinja2 and mustache
templates
 2025-11-19 16:52:32 -05:00
ccurme
b93d2f7f3a release(core): 0.3.79 (#33401) 2025-10-09 16:59:16 -04:00
ccurme
a763ebe86c release(anthropic): 0.3.22 (#33394) 2025-10-09 14:29:38 -04:00
ccurme
5fa1094451 fix(anthropic,standard-tests): carry over updates to v0.3 (#33393) 
Cherry pick of https://github.com/langchain-ai/langchain/pull/33390 and
https://github.com/langchain-ai/langchain/pull/33391.
 2025-10-09 14:25:34 -04:00
Anika
dd4de696b8 fix(core): handle parent/child mustache vars (#33346) 
Description:

currently mustache_schema("{{x.y}} {{x}}") will error. pr fixes

Issue: na
**Dependencies:**na

---------

Co-authored-by: Mason Daugherty <github@mdrxy.com>
 2025-10-09 08:52:10 -07:00
Mason Daugherty
5459ff1ee3 chore: cap lib upper bounds, run sync (#33320) 2025-10-06 17:39:01 -04:00
ccurme
f95669aa0a release(openai): 0.3.35 (#33299) 2025-10-06 10:51:01 -04:00
ccurme
3a465d635b feat(openai): enable stream_usage when using default base URL and client (#33296) 2025-10-06 10:22:36 -04:00
ccurme
0b51de4cab release(core): 0.3.78 (#33253) 2025-10-03 12:40:15 -04:00
ccurme
5904cbea89 feat(core): add optional include_id param to convert_to_openai_messages function (#33248) 2025-10-03 11:37:16 -04:00
Mason Daugherty
e16feb93b9 release(ollama): 0.3.10 (#33210) 2025-10-02 11:35:24 -04:00
Mason Daugherty
2bb57d45d2 fix(ollama): exclude None parameters from options dictionary (#33208) (#33209) 
fix #33206
 2025-10-02 11:32:21 -04:00
ccurme
d7cce2f469 feat(langchain_v1): update messages namespace (#33207) 2025-10-02 10:35:00 -04:00
Mason Daugherty
48b77752d0 release(ollama): 0.3.9 (#33200) 2025-10-01 22:31:20 -04:00
Mason Daugherty
6f2d16e6be refactor(ollama): simplify options handling (#33199) 
Fixes #32744

Don't restrict options; the client accepts any dict
 2025-10-01 21:58:12 -04:00
Mason Daugherty
a9eda18e1e refactor(ollama): clean up tests (#33198) 2025-10-01 21:52:01 -04:00
Mason Daugherty
a89c549cb0 feat(ollama): add basic auth support (#32328) 
support for URL authentication in the format
`https://user:password@host:port` for all LangChain Ollama clients.

Related to #32327 and #25055
 2025-10-01 20:46:37 -04:00
Sydney Runkle
a336afaecd feat(langchain): use decorators for jumps instead (#33179) 
The old `before_model_jump_to` classvar approach was quite clunky, this
is nicer imo and easier to document. Also moving from `jump_to` to
`can_jump_to` which is more idiomatic.

Before:

```py
class MyMiddleware(AgentMiddleware):
    before_model_jump_to: ClassVar[list[JumpTo]] = ["end"]

    def before_model(state, runtime) -> dict[str, Any]:
        return {"jump_to": "end"}
```

After

```py
class MyMiddleware(AgentMiddleware):

    @hook_config(can_jump_to=["end"])
    def before_model(state, runtime) -> dict[str, Any]:
        return {"jump_to": "end"}
```
 2025-10-01 16:49:27 -07:00
Sydney Runkle
a10e880c00 feat(langchain_v1): add async support for create_agent (#33175) 
This makes branching **much** more simple internally and helps greatly
w/ type safety for users. It just allows for one signature on hooks
instead of multiple.

Opened after https://github.com/langchain-ai/langchain/pull/33164
ballooned more than expected, w/ branching for:
* sync vs async
* runtime vs no runtime (this is self imposed)

**This also removes support for nodes w/o `runtime` in the signature.**
We can always go back and add support for nodes w/o `runtime`.

I think @christian-bromann's idea to re-export `runtime` from
langchain's agents might make sense due to the abundance of imports
here.

Check out the value of the change based on this diff:
https://github.com/langchain-ai/langchain/pull/33176
 2025-10-01 19:15:39 +00:00
Eugene Yurtsev
7b5e839be3 chore(langchain_v1): use list[str] for modifyModelRequest (#33166) 
Update model request to return tools by name. This will decrease the
odds of misusing the API.

We'll need to extend the type for built-in tools later.
 2025-10-01 14:46:19 -04:00
ccurme
740842485c fix(openai): bump min core version (#33188) 
Required for new tests added in
https://github.com/langchain-ai/langchain/pull/32541 and
https://github.com/langchain-ai/langchain/pull/33183.
 2025-10-01 11:01:15 -04:00
noeliecherrier
08bb74f148 fix(mistralai): handle HTTP errors in async embed documents (#33187) 
The async embed function does not properly handle HTTP errors.

For instance with large batches, Mistral AI returns `Too many inputs in
request, split into more batches.` in a 400 error.

This leads to a KeyError in `response.json()["data"]` l.288

This PR fixes the issue by:
- calling `response.raise_for_status()` before returning
- adding a retry similarly to what is done in the synchronous
counterpart `embed_documents`

I also added an integration test, but willing to move it to unit tests
if more relevant.
 2025-10-01 10:57:47 -04:00
ccurme
7d78ed9b53 release(standard-tests): 0.3.22 (#33186) 2025-10-01 10:39:17 -04:00
ccurme
7ccff656eb release(core): 0.3.77 (#33185) 2025-10-01 10:24:07 -04:00
ccurme
002d623f2d feat: (core, standard-tests) support PDF inputs in ToolMessages (#33183) 2025-10-01 10:16:16 -04:00
Mohammad Mohtashim
34f8031bd9 feat(langchain): Using Structured Response as Key in Output Schema for Middleware Agent (#33159) 
- **Description:** Changing the key from `response` to
`structured_response` for middleware agent to keep it sync with agent
without middleware. This a breaking change.
 - **Issue:** #33154
 2025-10-01 03:24:59 +00:00
Mason Daugherty
3e970506ba chore(core): remove runnable section from README.md (#33171) 2025-09-30 17:15:31 -04:00
ccurme
aac69839a9 release(openai): 0.3.34 (#33169) 2025-09-30 16:48:39 -04:00
ccurme
64141072a3 feat(openai): support openai sdk 2.0 (#33168) 2025-09-30 16:34:00 -04:00
Mason Daugherty
0795be2a04 docs(core): remove non-existent param from as_tool docstring (#33165) 2025-09-30 19:43:34 +00:00