Skip to content

deepagents-code==0.2.2

Choose a tag to compare

View all tags
@github-actions github-actions released this 26 Oct 20:39
deepagents-code==0.2.2
42e4f08
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

This update includes considerable security and performance improvements. We recommend upgrading as soon as feasible.

Features

  • Add support for GPT-6.1 Terra (#7295).
  • Bring selected text into the chat input (#6955).
  • Detect plugin updates and notify you when they are available (#7234).

Bug Fixes

  • Prevent project .env files from setting extension, shell, plugin, and skill trust (#7275).
  • Restrict links opened from the terminal UI to HTTP and HTTPS (#7277).
  • Restrict Git options in Auto mode's read-only fast path to an allowlist (#7276).
  • Stop the local server from uploading usage metadata to LangSmith (#7252).
  • Detect and repair enabled plugins with missing installed files (#7278).
  • Explain repeated skill-trust denials and clear them on /reload (#7291).
  • Fix transcript height calculations for folded tool rows and spacers so scrolling reaches the newest message (#7251, #7253).
  • Keep the terminal UI responsive when quitting during server startup (#7190).
  • Contain JavaScript subagent failures (#7211).
  • Manage sandbox lifetime from the client in non-interactive runs (#6368).
  • Preserve shipped bytecode for read-only installations (#7196).
  • Correctly read the current branch in reftable repositories through Git (#7268).

Performance Improvements

  • Reuse the parent's local context in forked subagents (#7233).
  • Show the terminal UI sooner by deferring imports and trace redacting-client setup until after the initial display (#7189, #7215).
  • Start the server earlier, before import prewarming, app-module imports, and post-display initialization (#7184, #7249, #7208).
  • Streamline server startup by deferring Textual imports, excluding custom routes from the OpenAPI specification, caching bytecode, and skipping a PyPI version check (#7242, #7193, #7198).
  • Build the agent graph during server startup and start runs without the in-memory queue's polling delay (#7214, #7243).
  • Improve streaming responsiveness by re-rendering Markdown only when blocks change, adapting the flush interval to rendering cost, and avoiding duplicate chunk merging (#7210, #7205, #7200).
  • Reduce transcript rendering overhead by compacting finished tool runs into lazy group rows, rendering completed replies as single widgets, limiting mounted rows to 350, and batching row removal (#7254, #7255, #7256, #7250).
  • Reduce background UI work by avoiding transcript relayouts on spinner ticks and DOM traversal on cache-expiry ticks (#7203, #7246).
  • Return control to the terminal sooner on exit by restoring it before server shutdown finishes, skipping the final garbage-collection pass, and avoiding the full shutdown wait for abandoned startups (#7192, #7191, #7194).
  • Speed up @ file-completion scoring (#7248).
  • Speed up the thread picker by filtering without remounting rows and keeping thread-list queries on a covering index (#7207, #7232).
  • Reduce post-turn data fetching to cost and cache channels, and cache SDK version lookups when building stream configuration (#7206, #7230).
  • Reduce tracing overhead by caching tracing secret redaction across runs, omitting BtwOperation hook inputs from traces, and skipping tracing-related thread hops when tracing is disabled (#7209, #7199, #7247).
  • Improve checkpoint efficiency by forwarding delta-channel history reads to the SQLite saver, queuing in-process writers instead of polling, and keeping the /btw system prompt out of checkpoints (#7244, #7245, #7202).
  • Skip redundant goal-notice history scans (#7201).

Thanks to our community contributors: Wei Wang (@wei-w-wang), Roydon Sequeira (@roydonsequeira)

Special thanks to everyone who reported the issues addressed in this release: Wei Wang (@wei-w-wang) (#6369), Roydon Sequeira (@roydonsequeira) (#7120)

Internal maintainers: Mason Daugherty (@mdrxy)

Released by: Mason Daugherty (@mdrxy)

Git log since deepagents-code==0.2.1

This commit history includes changes to this package. Commits are listed newest first.

  • 42e4f08 release(deepagents-code): 0.2.2 (#7182)
  • 315c2c9 feat(code): add GPT-6.1 Terra (#7295)
  • 3102bc0 chore(deps): bump certifi from 2026.10.19 to 2026.10.23 in /libs/code (#7287)
  • 50646c2 chore(deps): raise dependency minimums for all (#7293)
  • 47cf0fb fix(code): explain repeated skill-trust denials and clear them on /reload (#7291)
  • fa9966b release(langchain-modal): 0.0.7 (#7281)
  • ebd7592 feat(code): detect and notify plugin updates (#7234)
  • 4711bf2 fix(code): own sandbox lifetime from the client in non-interactive runs (#6368)
  • 89efee3 fix(code): flag and repair enabled plugins with missing installed files (#7278)
  • 71c9792 perf(code): shrink the mounted transcript window to 350 rows (#7256)
  • a18ca04 perf(code): compact finished live tool runs into lazy TOOL_GROUP rows (#7254)
  • f40ecaf fix(code): only open http and https links from the TUI (#7277)
  • 2abf323 fix(code): allowlist Git options in Auto mode's read-only fast path (#7276)
  • ebdee3e fix(code): stop project .env from setting extension, shell, plugin, and skill trust (#7275)
  • ca1b44e fix(code): read the branch of reftable repositories through git (#7268)
  • a8fa8e9 chore(deps): raise dependency minimums for all (#7272)
  • 9243a1c perf(code): render finished assistant replies as one widget (#7255)
  • f26eb54 chore(deps): bump filelock from 3.29.7 to 4.2.1 in /libs/code (#7257)
  • 793ac5a fix(code): give transcript spacers their real height so scrolling reaches the newest message (#7253)
  • 305a185 fix(code): stop the local server from uploading usage metadata to LangSmith (#7252)
  • 1a1d484 fix(code): correct transcript height hints for folded tool rows (#7251)
  • fb05a5d perf(code): skip traceable thread hops when tracing is off (#7247)
  • ec98477 perf(code): remove pruned transcript rows in one batch (#7250)
  • 2815f6b perf(code): drop dcode routes from the OpenAPI spec and defer Textual in the server (#7242)
  • e47f4e2 perf(code): start server runs without the in-memory queue's poll delay (#7243)
  • 97f518d perf(code): stop walking the DOM on the 1 Hz cache-expiry tick (#7246)
  • 89824bb perf(code): spawn the server before importing the app module (#7249)
  • 5fa7317 perf(code): forward delta-channel history reads to the SQLite saver (#7244)
  • 8d2c868 perf(code): reuse the parent's local context in forked subagents (#7233)
  • bca2b9c perf(code): speed up @ file completion scoring (#7248)
  • d5801f1 perf(code): queue in-process checkpoint writers instead of polling the ownership gate (#7245)
  • b82f3b3 perf(code): keep list_threads on its covering index (#7232)
  • d87a459 perf(code): cache the SDK version lookup in build_stream_config (#7230)
  • 6265589 chore(deps): raise dependency minimums for all (#7240)
  • a338166 chore(deps): bump setuptools from 84.1.0 to 84.2.1 in /libs/code (#7231)
  • 2adde3a chore(deps): bump setuptools from 83.0.0 to 84.1.0 in /libs/code (#7228)
  • 69e24a1 chore(deps): bump fastmcp from 4.0.3 to 4.1.2 in /libs/code (#7221)
  • db9b663 chore(deps): bump opentelemetry-api from 1.42.1 to 1.45.1 in /libs/code (#7220)
  • 94458c5 chore(deps): raise dependency minimums for all (#7219)
  • 470b075 perf(code): install the LangSmith redacting client after first paint (#7215)
  • b16a3c7 chore(deps): bump deepagents pin in deepagents-code to 0.7.32 (#7217)
  • 4d461e3 perf(code): build the agent graph during server startup (#7214)
  • 79bc54c chore(deps): raise dependency minimums for all (#7216)
  • e85c6b9 release(deepagents): 0.7.32 (#7188)
  • 798535d release(langchain-daytona): 0.0.9 (#7187)
  • 18f5543 perf(code): re-render streamed markdown only when blocks diverge (#7210)
  • 3d34b96 perf(code): filter the thread picker without remounting rows (#7207)
  • 4b2842c perf(code): start the server before post-paint init (#7208)
  • 5a3039b perf(code): cache LangSmith secret redaction across traced runs (#7209)
  • b93aff0 fix(code): contain JavaScript subagent failures (#7211)
  • 84b3bbe perf(code): stop relayouting the transcript on spinner ticks (#7203)
  • 2025b7c perf(code): keep the /btw system prompt out of checkpoints (#7202)
  • 4c91940 perf(code): merge streamed chunks once in the transcript recorder (#7200)
  • 5162474 perf(code): fetch only cost and cache channels after each turn (#7206)
  • 23ae4bb perf(code): omit BtwOperation hook inputs from traces (#7199)
  • 918a8c8 perf(code): skip redundant goal-notice history scans in GoalToolsMiddleware (#7201)
  • 61c9645 perf(code): skip the langgraph-api PyPI version check at server launch (#7198)
  • e79aaa5 perf(code): scale the stream flush interval to markdown render cost (#7205)
  • 2e875cd docs(code): add middleware trace and checkpoint state guidelines (#7197)
  • 76fe0cb fix(code): preserve shipped bytecode for read-only installs (#7196)
  • 1726e40 test(code): skip the non-UTF-8 filename test on macOS (#7195)
  • e53babb perf(code): let the server process cache its bytecode (#7193)
  • a7e09c1 perf(code): skip the full shutdown wait for an abandoned server startup (#7194)
  • 56971e2 perf(code): restore the terminal before the server finishes stopping (#7192)
  • a3b1ea9 perf(code): skip the final GC pass when the TUI exits (#7191)
  • 4a77435 fix(code): keep the TUI responsive when quitting during server startup (#7190)
  • a429412 perf(code): keep LangChain imports off the first-paint path (#7189)
  • 0f7ddb4 test(code): reset user LangSmith snapshot between tests (#7186)
  • 6d184fd perf(code): spawn server before waiting on import prewarm (#7184)
  • d54bcec chore(deps): require Daytona request timeout support (#7183)
  • 09b14d9 feat(code): bring selected text into the chat input (#6955)
  • 51f92b8 test(code): reset LangSmith launch snapshot between tests (#7180)

Contributors

wei-w-wang, mdrxy, and roydonsequeira
Assets 4
Loading