‹ Reports
The Dispatch

OSS Report: langchain-ai/langchain

LangChain Project Sees Steady Development with Emphasis on Documentation and Integration Enhancements

LangChain, a framework for building applications powered by large language models (LLMs), continues to see active development with a focus on improving documentation and expanding third-party integrations.

Recent Activity

Recent pull requests (PRs) indicate a strong emphasis on documentation enhancements and integration updates. Notable PRs include #25662, which adds documentation for ZhipuAIEmbeddings, and #25658, which improves AWS DynamoDB message history customization. The project also introduced new features like the Gaode weather querying tool (#25654) and enhanced prompt handling capabilities (#25674). Closed PRs such as #25673 and #25671 highlight efforts to fix typos and improve error messaging, respectively.

The development team is actively engaged in refining core functionalities and addressing community feedback. Team members such as Yahya Mouman, Eric Pinzur, and Ivan have been involved in enhancing integrations with Pinecone and CassandraGraphVectorStore, while others like William FH and Hasan Kumar focus on improving error handling and type safety. The team collaborates extensively, often co-authoring commits across various features.

Of Note

  1. Integration Focus: There is a clear trend towards enhancing integrations with third-party services like AWS, Neo4j, and Redis, reflecting LangChain's commitment to supporting diverse platforms.
  2. Documentation Efforts: A significant portion of recent activity is dedicated to improving documentation, which is crucial for user adoption and understanding of complex features.
  3. Draft PRs: Several draft PRs suggest ongoing development efforts that are not yet ready for final review, indicating a pipeline of upcoming features.
  4. Community Engagement: Active community involvement is evident from the diverse range of issues reported, covering bugs, documentation needs, and feature requests.
  5. Collaborative Development: The project exhibits a high level of collaboration among team members, often involving co-authorship on commits, which fosters a dynamic development environment.

Quantified Reports

Quantify Issues



Recent GitHub Issues Activity

Timespan Opened Closed Comments Labeled Milestones
7 Days 46 18 58 0 1
14 Days 96 32 121 1 1
30 Days 202 70 351 3 1
All Time 7428 6809 - - -

Like all software activity quantification, these numbers are imperfect but sometimes useful. Comments, Labels, and Milestones refer to those issues opened in the timespan in question.

Quantify commits



Quantified Commit Activity Over 30 Days

Developer Avatar Branches PRs Commits Files Changes
Bagatur 21 38/30/2 123 341 51482
Eugene Yurtsev 19 28/19/1 95 493 22327
Erick Friis 10 13/12/1 38 78 21170
Isaac Francisco 10 9/8/0 27 146 16006
ccurme 13 14/13/0 58 249 15959
John (Coniferish) 2 1/0/0 2 23 6234
Noah Peterson 1 0/0/0 1 1 6027
Scott Hurrey 1 4/2/2 2 38 4133
Yuki Watanabe 2 2/1/0 2 22 3856
Jacob Lee 8 3/2/0 14 37 2847
Casey Clements 3 0/0/0 3 22 2554
Brian Sam-Bodden 1 0/0/0 1 4 1817
Rajendra Kadam 2 2/2/0 3 7 1470
Leonid Ganeline 6 8/5/0 12 44 1198
maang-h 7 3/3/0 13 19 1143
Gareth 1 0/0/0 1 9 1018
Haijian Wang 1 0/0/0 1 9 869
Chaunte W. Lacewell 1 0/0/0 1 8 797
Serena Ruan (serena-ruan) 1 1/0/0 1 5 709
Anindyadeep 1 0/0/0 1 4 667
Lakshmi Peri 1 0/0/0 1 2 555
Vkzem 1 0/0/0 1 1 525
Bob Merkus 1 1/1/0 1 5 453
Rave Harpaz 1 0/0/0 1 3 426
yonarw 1 0/0/0 1 5 410
William FH 2 1/1/0 4 11 387
ChengZi 1 0/0/0 1 2 377
Vadym Barda (vbarda) 1 1/0/0 2 3 338
Tomaz Bratanic 2 0/0/0 3 3 314
Vishnu Nandakumar 1 0/0/0 1 6 306
Anush 4 0/0/0 4 6 283
David Gao 1 0/0/0 1 2 277
Aayush Kataria 1 0/0/0 1 4 264
Oleg Kulyk 1 0/0/0 1 4 260
Yunus Emre Özdemir 1 0/0/0 1 1 248
thedavgar 1 0/0/0 1 1 225
Mohammad Mohtashim 1 0/0/0 2 2 206
mschoenb97IL 2 1/1/0 2 2 196
남광우 1 0/0/0 1 2 185
ZhangShenao 6 5/2/1 11 28 182
Daniel Glogowski 1 0/0/0 1 1 174
Nuno Campos 2 0/0/0 2 4 166
Harrison Chase 1 1/1/0 1 5 161
Chengyu Yan 1 0/0/0 1 2 149
nobbbbby 1 0/0/0 1 2 148
Tamir Zitman 1 0/0/0 1 4 124
Fernando de Oliveira 1 0/0/0 3 2 118
Igor Drozdov 1 0/0/0 1 4 114
WU LIFU 1 0/0/0 1 4 110
Lance Martin 1 0/0/0 1 1 103
Joel Akeret 1 0/0/0 1 2 87
Maddy Adams 1 0/0/0 1 1 86
Christophe Bornet 1 6/2/0 2 3 75
KyrianC 1 0/0/0 1 10 69
gbaian10 2 4/3/0 7 15 68
Hasan Kumar 1 0/0/0 1 3 64
rick-SOPTIM 1 0/0/0 1 2 56
Pavel (PabloKarpacho) 1 1/0/0 1 2 55
Sunish Sheth 1 0/0/0 1 1 53
Mateusz Szewczyk 1 0/0/0 1 2 50
Dobiichi-Origami 1 0/0/0 1 1 48
Gram Liu 1 0/0/0 3 2 40
Stuart Marsh 1 0/0/0 1 2 38
Cristi Burcă 1 0/0/0 1 3 30
Jerron Lim 2 0/0/0 3 2 30
Eric Pinzur 1 1/1/0 1 1 29
sslee 1 1/1/0 1 3 26
Ben Chambers 3 0/0/0 3 3 25
Chen Xiabin 1 1/1/0 1 1 24
Abraham Omorogbe 1 1/1/0 1 1 22
Hayden Wolff 1 0/0/0 1 2 22
Ivan 1 2/1/1 1 2 22
cffranco94 1 0/0/0 1 1 21
BottlePumpkin 1 0/0/0 1 1 21
Diverrez morgan 1 0/0/0 1 1 21
Swastik-Swarup-Dash 1 1/1/0 1 1 20
Noah Mayerhofer 1 1/1/0 1 1 17
Dylan 1 1/1/0 1 1 16
Shailendra Mishra 1 0/0/0 1 1 16
Krishna Kulkarni 1 1/1/0 1 2 15
Hassan El Mghari 1 1/1/0 1 3 14
David Robertson 1 0/0/0 1 1 14
sykp241095 2 0/0/0 2 3 14
Chengzu Ou 1 1/1/0 1 1 13
Hassan Memon 1 0/0/0 2 1 13
Marc Gibbons 1 0/0/0 1 1 11
Shivendra Soni 1 0/0/0 1 1 11
Alex Sherstinsky 1 0/0/0 1 1 11
Thin red line 未来产品经理 1 1/1/0 1 2 10
Flávio Knob 1 2/2/0 2 1 8
ololand 1 0/0/0 1 1 8
monysun 1 0/0/0 1 1 7
Jim Baldwin 1 0/0/0 1 1 7
Jabir 1 1/1/0 1 1 6
Jesse Wright 1 0/0/0 1 1 6
JasonJ 1 0/0/0 1 3 6
Luke 1 1/1/0 1 1 6
Daniel Campos 1 0/0/0 1 1 6
AmosDinh 1 0/0/0 1 1 5
yahya-mouman 1 1/1/0 1 1 5
Dristy Srivastava 1 2/1/0 1 1 5
wangda 1 1/1/0 1 2 4
Chad Juliano 1 0/0/0 1 2 4
Carlos André Antunes 1 0/0/0 1 1 4
Luca Dorigo 1 0/0/0 1 1 4
orkhank 1 0/0/0 1 1 4
alexqiao 1 0/0/0 1 1 4
Ikko Eltociear Ashimine 2 1/1/0 2 2 4
Mikhail Khludnev 1 2/1/0 1 1 4
Brace Sproul 1 0/0/0 1 2 4
jakerachleff 1 1/1/0 1 1 4
thiswillbeyourgithub 1 0/0/0 1 1 4
Nik Jmaeff 1 0/0/0 1 1 3
Maurits Bos 1 1/1/0 1 1 3
Cong Peng 1 0/0/0 1 1 2
stydxm 1 0/0/0 1 1 2
Ken Jenney 1 0/0/0 1 1 2
Chuck Wooters 1 0/0/0 1 1 2
blaufink 1 0/0/0 1 1 2
Matt Kandler 1 1/1/0 1 1 2
ogawa 1 0/0/0 1 1 2
Werner van der Merwe 1 1/1/0 1 1 2
Alexey Lapin 1 0/0/0 1 1 2
blueoom 1 0/0/0 1 1 2
cüre 1 0/0/0 1 1 2
Morteza Hosseini 1 0/0/0 1 1 2
Soichi Sumi 1 0/0/0 1 1 2
Francisco Kurucz 1 0/0/0 1 1 2
Moritz 1 0/0/0 1 1 2
Anneli Samuel 1 0/0/0 1 1 2
Aryan Singh 1 0/0/0 1 1 2
jigsawlabs-student 1 0/0/0 1 1 2
Fedor Nikolaev 1 1/1/0 1 1 1
Philippe PRADOS 1 0/0/0 1 1 1
Michal Gregor 1 0/0/0 1 1 1
Geovanny Fajardo (geosp) 0 1/0/0 0 0 0
Alejandro Companioni (acompa) 0 1/0/0 0 0 0
Sanjay Parajuli (xanjay) 0 2/0/1 0 0 0
None (Huarong) 0 1/0/1 0 0 0
Martin Triska (MacanPN) 0 1/0/0 0 0 0
None (zysoong) 0 1/0/0 0 0 0
None (amirai21) 0 1/0/0 0 0 0
Erik Lindgren (erikml-db) 0 1/0/0 0 0 0
Jakub W. (jakubikan) 0 1/0/0 0 0 0
Karthik Sunil (karthikcs) 0 1/0/0 0 0 0
Lucien (LucienShui) 0 1/0/0 0 0 0
Ali Altıparmak (ali6parmak) 0 1/0/0 0 0 0
Niels Bantilan (cosmicBboy) 0 1/0/0 0 0 0
tim-watcha (tim-watcha) 0 1/0/0 0 0 0
None (olgamurraft) 0 1/0/0 0 0 0
clement.l (Doge-is-Dope) 0 1/0/0 0 0 0
hpx (hpx502766238) 0 1/0/1 0 0 0
Parsa Abbasi (parsa-abbasi) 0 1/0/0 0 0 0
None (venkatram-dev) 0 1/0/0 0 0 0
None (Tyler-Odenthal) 0 1/0/0 0 0 0
None (XinlingWang0628) 0 1/0/0 0 0 0
Dima Matasov (mrdrprofuroboros) 0 1/0/0 0 0 0
Karthik Bharadhwaj (karthikbharadhwajKB) 0 1/0/0 0 0 0

PRs: created by that dev and opened/merged/closed-unmerged during the period

Detailed Reports

Report On: Fetch issues



Recent Activity Analysis

Recent GitHub issue activity for the LangChain project shows a diverse range of issues being reported and discussed. The issues cover various aspects of the project, including bugs, documentation improvements, feature requests, and integration challenges with third-party tools and platforms. Notably, there are several issues related to specific integrations such as OpenAI, Hugging Face, Redis, and others, indicating active use and testing of these components by the community.

A recurring theme among the issues is the need for better documentation and examples, particularly for new features or complex integrations. There are also several reports of bugs or unexpected behavior when using specific tools or configurations, suggesting areas where the project could improve its robustness or error handling.

Issue Details

  1. #25670: A documentation issue regarding cache integrations was created 0 days ago. It highlights the need for better organization of cache integration docs.
  2. #25664: A bug report about LangChain dumps not being implemented correctly was created 0 days ago. This issue is labeled as affecting core functionalities.
  3. #25659: A bug in SelfQueryRetriever interpreting numbers as strings was reported 0 days ago. This issue is marked for investigation.
  4. #25651: A bug related to Chroma implementation was reported 1 day ago, indicating ongoing challenges with this integration.
  5. #25629: A documentation mismatch in DuckDuckGo tool's max_results parameter was reported 1 day ago.
  6. #25613: A bug in Redis cache causing ValidationError on retrieval was reported 1 day ago.
  7. #25609: A bug in Langchain document loader related to an upstream nltk issue was reported 1 day ago.
  8. #25606: An issue with inserting GraphDocuments into Memgraph was reported 1 day ago.
  9. #25603: A logic error when nesting Chain without error messages was reported 2 days ago.
  10. #25575: A bug in YandexGPT client returning an error when invoked was reported 2 days ago.

These issues reflect active engagement from the community in identifying and addressing problems within the LangChain framework, contributing to its ongoing development and improvement.

Report On: Fetch pull requests



Overview

The provided data consists of a comprehensive list of open and closed pull requests (PRs) for the LangChain project, hosted on GitHub under the repository langchain-ai/langchain. The project is a framework designed to facilitate the development of applications powered by large language models (LLMs). It has gained significant traction with a large number of stars and forks, indicating strong community interest. The PRs cover a wide range of updates, including bug fixes, feature enhancements, documentation improvements, and integration updates.

Summary of Pull Requests

Recent Open Pull Requests

  1. #25674: Introduces support for passing dictionaries as templates to chat prompt templates. This PR is currently in draft status and aims to simplify the process of updating existing prompt template classes.
  2. #25662: Adds documentation for ZhipuAIEmbeddings, using an integration doc template.
  3. #25658: Improves the DynamoDB message history by adding a history_messages_key, allowing users to customize the key used for chat history.
  4. #25654: Adds a new tool for querying Gaode weather, along with relevant documentation and API key requirements.
  5. #25652: Enhances the UCFunctionToolkit by supporting additional parameters when executing functions.

Recent Closed Pull Requests

  1. #25673: Fixes a typo in various notebooks related to multi-modal RAG and langgraph self-RAG.
  2. #25672: Releases version 0.0.30 of the CLI.
  3. #25671: Improves error messaging for DeepInfra 500 errors by exposing more context from the error response.
  4. #25669: Adds docstrings for tool calling and structured output in ChatZhipuAI.
  5. #25665: Updates the README to improve understanding of LangChain Expression Language (LCEL).

Analysis of Pull Requests

The LangChain project is actively maintained with a high volume of pull requests addressing various aspects of the framework. A significant portion of the PRs focuses on enhancing documentation, which is crucial for user adoption and understanding of complex features like embeddings and prompt templates.

There is a clear emphasis on improving integrations with third-party services such as AWS, Pinecone, and Neo4j, reflecting LangChain's commitment to providing robust support for various platforms used in LLM applications. This is evident in PRs like #25658 and #25660, which enhance AWS DynamoDB integration and add retry logic for Neo4j sessions, respectively.

The project also shows an ongoing effort to refine its core functionalities, as seen in PRs like #25674 and #25652, which aim to simplify prompt handling and function execution within the framework.

Anomalies include some PRs that were closed without merging, such as #25584, which indicates potential issues or reconsiderations during the review process. Additionally, there are several draft PRs that suggest ongoing development efforts that are not yet ready for final review.

Overall, LangChain's active development cycle is characterized by frequent updates that enhance both functionality and usability. The project's maintainers appear responsive to community feedback and are continuously working to integrate new features while maintaining robust documentation. This approach not only fosters community engagement but also ensures that LangChain remains a competitive option for developers building LLM-powered applications.

Report On: Fetch commits



Development Team and Recent Activity

Team Members and Activities

  1. Yahya Mouman:

    • Worked on adding the ID field to similarity document results in the Pinecone integration.
    • Collaborated with Chester Curme.

  2. Eric Pinzur:

    • Added metadata filtering support to CassandraGraphVectorStore.
    • Fixed type conversion issues highlighted by mypy.
    • Collaborated with Chester Curme.

  3. Ivan (istrebitel-1):

    • Fixed UnionType type variable replacement in langchain_core.
    • Collaborated with Chester Curme.

  4. William FH (hinthornw):

    • Improved error messages and added tests for RunnableWithMessageHistory.
    • Collaborated with Eugene Yurtsev.

  5. Hasan Kumar (mintuhouse):

    • Fixed type warnings when passing Runnable as an agent to AgentExecutor.
    • Collaborated with Chester Curme.

  6. Leonid Ganeline (leo-gan):

    • Updated documentation references for integrations and added new provider pages.

  7. Maurits Bos (MbBrainz):

    • Updated pyproject.toml for openai-functions-agent-gmail package to fix ModuleOrPackageNotFound error.

  8. Erick Friis (efriis):

    • Released CLI version 0.0.30.
    • Made several updates across different branches, including chroma and unstructured packages.

  9. Dylan Griffith:

    • Updated GitLab notebook terminology from "Pull Request" to "Merge Request".

  10. Brian Sam-Bodden (bsbodden):

    • Added Redis Partner Package notebooks and updated related documentation.

  11. Rajendra Kadam (Raj725):

    • Refactored PebbloRetrievalQA and PebbloSafeLoader.
    • Improved code readability and removed unused code.

  12. Krishna Kulkarni (kk9393):

    • Added a doc_limit parameter for limiting recent documents fetched from MongoDB database.

  13. Chester Curme:

    • Co-authored multiple commits across various features and bug fixes, collaborating with several team members.

  14. Noah Mayerhofer (noahmay):

    • Added retry logic for session expired exceptions in neo4j integration.

  15. Yuki Watanabe (B-Step62):

    • Added ChatDatabricks implementation and partner package directory for Databricks.
    • Collaborated with Erick Friis.

  16. Scott Hurrey (shurrey):

    • Added BoxRetriever for langchain_box and updated related tests and documentation.
    • Collaborated with Erick Friis.

  17. Bagatur Baskaryan (baskaryan):

    • Made numerous contributions across various branches, including updates to core, partners, and experimental packages.
    • Worked on standard testing, API reference updates, and integration improvements.

  18. Eugene Yurtsev (eyurtsev):

    • Contributed extensively to core improvements, integration updates, and testing enhancements.
    • Worked on pydantic compatibility and rate limiter implementations.

  19. Bob Merkus:

    • Updated Ollama documentation to reflect recent changes in tool calling support.

  20. Jabir (ja6ir):

    • Updated azuresearch.py to handle complex metadata types without errors.

  21. Abraham Omorogbe (AbeOmor):

    • Added vector store and memory usage documentation for Azure Database for PostgreSQL.

  22. Mohammad Mohtashim (keenborder786):

    • Fixed OpenAI WhisperParser issues related to text object returns.

  23. Thin red line 未来产品经理 (jiru1997):

    • Addressed prompt validation issues in create_stuff_documents_chain function.

  24. Hassan El Mghari (Nutlope):

    • Updated Together base URL from .ai to .xyz.

  25. Gbaian10:

    • Made several documentation updates, including installation prompts for dependencies like beautifulsoup4 and nltk.

  26. Chengzu Ou (freemso):

    • Fixed demo notebook issue in Databricks Vector Search section.

  27. Mikhail Khludnev (mkhludnev):

    • Corrected KeyError issue in qdrant.ipynb by updating method calls.

  28. Dristy Srivastava (dristysrivastava):

    • Updated SharePoint loader metadata handling to include full path URLs.

  29. ZhangShenao:

    • Improved API documentation for PineconeVectorStore and other components.

  30. Wangda (daziz):

    • Corrected spelling mistakes in provider documentation files.

  31. Isaac Francisco (isahers1):

    • Made extensive contributions across multiple branches, focusing on embedding tests, documentation standardization, and integration improvements.

  32. Bob Merkus: Updated Ollama documentation to reflect recent changes in tool calling support.

Patterns, Themes, and Conclusions

  • The LangChain project exhibits a high level of collaboration among team members, often involving co-authorship on commits.
  • There is a strong emphasis on improving documentation, with multiple contributors focusing on updating references, correcting terminology, and adding new provider pages.
  • The development team actively addresses bug fixes across various integrations, such as Pinecone, CassandraGraphVectorStore, MongoDB, Neo4j, AzureSearch, etc., indicating a commitment to maintaining robust functionality across the framework's components.
  • There is ongoing work on enhancing the framework's core capabilities, including improvements to error handling, type safety, rate limiting, and compatibility with newer versions of dependencies like pydantic.
  • The project shows a consistent effort in expanding its integration capabilities by adding support for new tools and platforms like Databricks, BoxRetriever, Redis Partner Package, etc., reflecting an adaptive approach to evolving user needs.
  • Overall, LangChain demonstrates active development with frequent updates that enhance both its core functionality and its extensibility through integrations with third-party services.