LangChain Deep Agents new version release
The story so far
LangChain Deep Agents are gradually releasing new versions, fixing multiple bugs and adding new features. Version 0.7.29 fixed bugs such as handling trailing slashes in routing for `CompositeBackend`, rejecting binary file text editing, and skipping binary files in in-memory `grep`. Version 0.7.30 supports asynchronous sub-agent queues to follow up on tasks without interrupting current work, and fixed bugs such as asynchronous sub-agent check report issues during server cancellation, thread read failures, removing associated reasoning blocks from summaries, and local Shell backend output limits. Version 0.7.31 fixed bugs such as omitting asynchronous task IDs from user responses by default and retaining filesystem permissions in SDK middleware overrides, and added a feature to expose execution output metadata in tracking. Version 0.7.32 improved performance by avoiding loading `langchain_anthropic` and `langchain_openai` modules during SDK import. Version 0.7.33 defaults to erroring out when models reject and reporting sub-agent rejections, adds asynchronous sub-agent notification metadata, logs sub-agent checkpoint task call IDs, and fixed issues such as Python `grep` fallback limiting file ranges, truncating long result lines, and erroring out on compression failures, while officially formalizing the rubric middleware. Version 0.7.34 fixed summary truncation issues by anchoring message IDs. Version 0.7.35 fixed bugs such as retaining parent agent prompt cache prefixes for sub-agents, preserving symlink names in virtual filesystem results, and marking `read_file` media and Nemotron nudge as harness notifications. Version 0.7.37 fixed issues with optional prompt cache lookup and updated minimum dependency version requirements.
AI-written from the reports · Updated yesterday
Developments
8 developments
- 13 Nov, 04:03 · 1 reportsdeepagents==0.7.37LangChain Deep Agents releases:LangChain Deep Agents 0.7.37 released
- 10 Nov, 07:40 · 1 reportsdeepagents==0.7.35 releaseLangChain Deep Agents releases:LangChain Deep Agents 0.7.35 released
- 6 Nov, 05:15 · 1 reportsdeepagents==0.7.34LangChain Deep Agents releases:LangChain Deep Agents 0.7.34 released
- 6 Nov, 04:40 · 1 reportsLangChain Deep Agents v0.7.33 releasedLangChain Deep Agents releases:LangChain Deep Agents v0.7.33 released
- 24 Oct, 05:30 · 1 reportsdeepagents==0.7.32LangChain Deep Agents releases:LangChain Deep Agents 0.7.32 released: improving SDK performance optimization
- 24 Oct, 00:27 · 1 reportsdeepagents==0.7.31LangChain Deep Agents releases:LangChain Deep Agents 0.7.31 released
- 22 Oct, 10:27 · 1 reportsdeepagents==0.7.30LangChain Deep Agents releases:LangChain Deep Agents v0.7.30 released
- 21 Oct, 00:17 · 1 reportsdeepagents==0.7.29LangChain Deep Agents releases:LangChain Deep Agents 0.7.29 released
Coverage timeline
Follow the coverage to see every side of the story.
- LangChain Deep Agents releasesLangChain Deep Agents 0.7.37 released
LangChain Deep Agents SDK has fixed the lookup cache issue for optional prompt word caching and updated the minimum version requirements for dependencies.
- LangChain Deep Agents releasesLangChain Deep Agents 0.7.35 released
LangChain Deep Agents 0.7.35 released, fixing bugs such as preserving the parent agent's prompt cache prefix for child agents, retaining symbolic link names in virtual file system results, marking `read_file` media and Nemotron nudge as harness notifications, etc.
- LangChain Deep Agents releasesLangChain Deep Agents 0.7.34 released
LangChain Deep Agents 0.7.34 released, fixed the summary truncation handling issue, and addressed the problem by anchoring the message ID.
- LangChain Deep Agents releasesLangChain Deep Agents v0.7.33 released
LangChain Deep Agents v0.7.33 released, defaulting to error reporting when the model refuses and reporting sub-agent refusal, adding asynchronous sub-agent notification metadata, recording sub-agent checkpoint task call ID. Fixed issues with Python `grep` fallback file range limitation, long result line truncation, and compression failure error reporting, and officially formalized the rubric middleware.
- LangChain Deep Agents releasesLangChain Deep Agents 0.7.32 released: improving SDK performance optimization
LangChain Deep Agents 0.7.32 released, improves performance by avoiding loading `langchain_anthropic` and `langchain_openai` modules during SDK import. This update optimizes SDK import speed, with specific optimization details recorded in the Git log.
- LangChain Deep Agents releasesLangChain Deep Agents 0.7.31 released
LangChain Deep Agents 0.7.31 released, fixed bugs such as omitting asynchronous task IDs from user responses by default, retaining filesystem permissions in SDK middleware overrides, and added a feature to expose execution output metadata retention in tracking.
- LangChain Deep Agents releasesLangChain Deep Agents v0.7.30 released
LangChain Deep Agents v0.7.30 released, supporting asynchronous sub-agent queues to follow up on tasks without interrupting current work, fixes issues such as asynchronous sub-agent check report problems during server cancellation, thread read failure retries, removal of summary-related reasoning blocks, and local Shell backend output limits
- LangChain Deep Agents releasesLangChain Deep Agents 0.7.29 released
LangChain Deep Agents 0.7.29 released, fixes bugs in `CompositeBackend` handling trailing slash routes, rejecting binary file text editing, in-memory `grep` skipping binary files, etc.
Heat over time
- Now
- 3
- Peak
- 1010 Nov, 08:00
- 24h change
- -51%
Hover or tap the chart to see each hour; use the arrow keys on a keyboard.