Skip to content

Release history

deepagents releases

Agent harness built with LangChain and LangGraph. Equipped with a planning tool, a filesystem backend, and the ability to spawn subagents - well-equipped to handle complex agentic tasks.

Back to tool Latest release

All releases

138 shown

No immediate action
langchain-quickjs==0.1.4 Bug fix

Middleware configuration fix

Open
No immediate action
langchain-modal==0.0.5 Maintenance

Routine maintenance and dependency updates.

Open
No immediate action
langchain-runloop==0.0.6 New feature

Blueprint bootstrapping

Open
No immediate action
langchain-daytona==0.0.7 Maintenance

Routine maintenance and dependency updates.

Open
No immediate action
deepagents-code==0.1.9 Maintenance

Routine maintenance and dependency updates.

Open
No immediate action
deepagents==0.6.8 Breaking risk

Dropped subagent exports

Open
Review required
deepagents-acp==0.0.7 Maintenance
Dependencies

Routine maintenance and dependency updates.

Open
No immediate action
deepagents-code==0.1.8 Mixed

MCP error modal + recovery commands

Open
deepagents-cli==0.2.0 Breaking risk
Breaking changes
  • `deepagents deploy` now targets the Managed Deep Agents API; `deepagents dev`, `deepagents.toml`, and the `--config`/`--dry-run` flags are removed. Run `deepagents init` to re-scaffold (`agent.json` + `tools.json`).
Notable features
  • Migrate `deepagents deploy` to use Managed Deep Agents API
Full changelog

⚠ BREAKING CHANGES

  • deepagents deploy now targets the Managed Deep Agents API; deepagents dev, deepagents.toml, and the --config/--dry-run flags are removed. Run deepagents init to re-scaffold (agent.json + tools.json). See PR for the full migration guide.

Features

  • Migrate deepagents deploy to use Managed Deep Agents API (#3609) (72aef52)

Bug Fixes

  • Stable HumanMessage IDs across resumed threads (#3591) (82c3194)

Internal maintainers: @sydney-runkle, @mdrxy, @victorm-lc

View release on GitHub
No immediate action
langchain-quickjs==0.1.3 New feature

REPL persistence + swarm tool

Open
No immediate action
deepagents-code==0.1.7 Bug fix

MCP rendering crash fix

Open
No immediate action
deepagents==0.6.7 Bug fix

Propagate goto and graph

Open
No immediate action
deepagents==0.6.6 New feature

Pass state_schema to create_deep_agent

Open
No immediate action
deepagents==0.6.5 New feature

RubricMiddleware + grep fallback logging

Open
No immediate action
deepagents-code==0.1.6 New feature

Install extras + MCP reconnect

Open
No immediate action
deepagents-code==0.1.5 Bug fix

aiosqlite thread hang fix

Open
No immediate action
deepagents==0.6.4 Bug fix

Grep stability + HumanMessage ID stability

Open
No immediate action
deepagents-code==0.1.4 New feature

--sandbox-snapshot-name + mcp config unification + quickJS interpreter

Open
No immediate action
deepagents-cli==0.1.2 Bug fix

Env var expansion in mcp.json

Open
No immediate action
deepagents==0.6.3 Bug fix

UUIDs for ID-less messages

Open
No immediate action
deepagents-code==0.1.2 New feature

/model toggle

Open
No immediate action
deepagents==0.6.2 New feature

system_prompt override + bug fixes

Open
No immediate action
deepagents-cli==0.1.1 New feature

`--host` option

Open
Upgrade now
deepagents-code==0.1.1 Bug fix
RCE / SSRF

SSRF guard + LangSmith fix

Open
No immediate action
deepagents-code==0.1.0 Maintenance

Routine maintenance and dependency updates.

Open
No immediate action
deepagents==0.6.1 Bug fix

SDK profile fix

Open
No immediate action
langchain-modal==0.0.4 Maintenance

Routine maintenance and dependency updates.

Open
No immediate action
langchain-runloop==0.0.5 Maintenance

Routine maintenance and dependency updates.

Open
No immediate action
langchain-daytona==0.0.6 Maintenance

Routine maintenance and dependency updates.

Open
Review required
deepagents==0.6.0 New feature
RCE / SSRF Dependencies

CodeInterpreterMiddleware + v3 event streaming

Open
No immediate action
deepagents-cli==0.0.57 New feature

OSC 9;4 progress helper

Open
No immediate action
langchain-quickjs==0.1.2 Breaking risk

Middleware rename

Open
deepagents-cli==0.0.55 Bug fix

Fixed the Thinking spinner not showing during `edit_file` approval.

Full changelog

Bug Fixes

  • Keep Thinking spinner during edit_file approval (#3310) (d06133f)

Internal maintainers: @mdrxy

View release on GitHub
deepagents-cli==0.0.54 New feature
Notable features
  • Allow filtering threads by directory
  • Show sandbox provider in header `sub_title`
  • Surface sandbox startup errors in server crash messages
Full changelog

Features

  • Allow filtering threads by directory (#3290) (07a973a)
  • Show sandbox provider in header sub_title (#3295) (779f7f4)
  • Surface sandbox startup errors in server crash messages (#3306) (2606d57)

Bug Fixes

  • Allow TUI launch without preconfigured model credentials (#3301) (aa28ff7)
  • Suppress token placeholder on first invoke (#3302) (c21dee2)
  • Sync thread ID and render clickable link on /clear (#3305) (0904974)

Thanks to our community contributors: @Chesars, @yuvrxj-afk

Internal maintainers: @mdrxy

View release on GitHub
deepagents==0.5.9 Bug fix

Fixed import of profile symbols by referencing `harness_profiles`.

Full changelog

Bug Fixes

  • Import profile symbols directly from harness_profiles (#3291) (503453c)

Internal maintainers: @mdrxy

View release on GitHub
deepagents-cli==0.0.52 New feature
Notable features
  • Launch theme, splash tips/version visibility, and TUI footer visibility controlled via new env vars
  • Emit `user.name.set` hook during onboarding
  • External event listener via Unix domain socket
Full changelog

Features

  • New environment variables
  • Emit user.name.set hook during onboarding (#3177) (43a494e)
  • External event listener via Unix domain socket (#3207) (ffeac22)
  • Strip provider-specific model prefixes in status bar footer (#3222) (eaa1d82)
  • (Deploy) ContextHubBackend for iteration in deepagents-deploy (#2923) (fb0f0ba)
  • (Deploy) Auto-wire issues board with context hub repo handle (#3231) (94628fe)

Bug Fixes

  • Allow multiple-choice labels to wrap in narrow ask-user menus (#3257) (79a38d9)
  • Catch WorkerCancelled in await_prewarm_imports (#3178) (aa0d0f7)
  • Halt turn immediately when ask_user is cancelled (#3258) (00d9b99)
  • Only restore iTerm2 cursor guide when it was enabled at startup (#3256) (2620ff5)
  • Pause input cursor blink when terminal loses OS focus (#3246) (8582b83)
  • Scroll tall ask_user prompts to top of viewport (#3238) (705daa7)
  • Suppress noisy network warnings from interrupt cleanup (#3250) (4e4a8fe)

Thanks to our community contributors: @yuvrxj-afk

Internal maintainers: @vishnu-ssuresh, @mdrxy

View release on GitHub
deepagents==0.5.8 Breaking risk

Fixed deprecated‑use warnings in CompositeBackend path mutation.

Full changelog

Bug Fixes

  • Avoid deprecated-use warnings in CompositeBackend path mutation (#3244) (64d45f6)

Internal maintainers: @mdrxy

View release on GitHub
langchain-quickjs==0.1.1 New feature patches CVE-2026-42208
Notable features
  • quickjs: propagate return types
Full changelog

Features

Reverted Changes

Internal maintainers: @hntrl, @ccurme

View release on GitHub
deepagents-cli==0.0.51 New feature
Notable features
  • /reload skill now produces a diff report of changes
  • MCP config discovery paths and format displayed in help screens
Full changelog

Features

  • /reload skill diff report (#3153) (7c3278a)
  • Surface MCP config discovery paths and format in help screens (#3152) (ce6887c)

Internal maintainers: @mdrxy

View release on GitHub
deepagents==0.5.7 Bug fix

Auto-added GP subagents now inherit parent permissions.

Full changelog

Bug Fixes

  • Auto-added GP subagent inherits parent permissions (#3131) (0d55b3b)
  • Default OpenRouter routing to ignore Azure upstream (#3157) (01a9113)

Internal maintainers: @sydney-runkle, @jacobkleiman-LC, @eyurtsev, @mdrxy

View release on GitHub
langchain-quickjs==0.1.0 Breaking risk
Breaking changes
  • Replaces the previous interpreter path with a new QuickJS runtime implementation backed by quickjs-rs.
Notable features
  • Added `max_ptc_calls` budget for ptc calls
  • Implemented snapshot-based REPL persistence between turns
  • Surface tool exceptions as the original error
Full changelog

This release introduces a new QuickJS runtime implementation backed by
quickjs-rs, replacing the previous interpreter path.

Features

  • quickjs: add max_ptc_calls budget for ptc calls (#2994) (13f6c2d)
  • quickjs: add snapshot-based repl persistence between turns (#3064) (c46feed)
  • quickjs: surface tool exceptions as the original error (#3049) (d96dc8c)
  • quickjs: use quickjs-rs (#2979) (7491899)

Bug Fixes

  • quickjs: add ls_code_input_language metadata to eval tool (#3062) (b9bc674)
  • quickjs: bound console buffering at capture time (#2999) (251e405)
  • quickjs: handle top-level await in snapshot step (#3161) (b330c22)
  • quickjs: keep PTC loop and runtime context in _PTCState (#3134) (c70cc5a)
  • quickjs: per-thread_id Runtimes (#2931) (4021b03)
  • quickjs: prompt improvements (#2564) (4999c6b)
  • quickjs: remove ptc command buffering (#3023) (ac1218a)

Changelog

Prior Releases

Versions prior to 0.0.2 were released without release-please and do not have changelog entries. Refer to the releases page for details on previous versions.

Internal maintainers: @sydney-runkle, @vishnu-ssuresh, @hntrl, @jkennedyvz, @ccurme, @eyurtsev, @mdrxy

View release on GitHub
deepagents-cli==0.0.50 Bug fix

Fixed display of detached HEAD commit in local context.

Full changelog

Bug Fixes

  • Show detached HEAD commit in local context (#3144) (569d4f0)

Internal maintainers: @mdrxy

View release on GitHub
deepagents-cli==0.0.49 New feature
Notable features
  • First-run onboarding flow with in-TUI API key entry via /auth
  • Set-as-default selection in /agents picker
Full changelog

Features

Bug Fixes

Internal maintainers: @mdrxy

View release on GitHub
deepagents-cli==0.0.48 Bug fix

Fixed --model-params not applied when selecting a model via /model.

Full changelog

Bug Fixes

  • Apply --model-params on /model re-select (#3072) (50546dd)

Internal maintainers: @mdrxy

View release on GitHub
deepagents==0.5.6 Bug fix
Notable features
  • Added write preflight and native read capabilities to langsmith sandbox
Full changelog

Bug Fixes

  • Add write preflight and native read to langsmith sandbox (#2695) (741221c)
  • Propagate CompiledSubAgent name into lc_agent_name metadata (#3045) (f671e6b)

Thanks to our community contributors: @yuvrxj-afk

Internal maintainers: @mdrxy

View release on GitHub
deepagents-cli==0.0.47 Mixed
Notable features
  • Auto-discover Textual built-in themes
Full changelog

Features

Bug Fixes

  • Recover from failed server startup via /model (#3056) (6ee375e)

Internal maintainers: @mdrxy

View release on GitHub
deepagents-cli==0.0.46 New feature
Notable features
  • Reworked MCP integration with added OAuth login capability
  • `allowedTools` and `disabledTools` filters introduced in MCP config
Full changelog

Features

  • Rework MCP integration and add OAuth login (#2906) (6b7b0be)
  • allowedTools and disabledTools filters in MCP config (#2835) (eafd691)

Internal maintainers: @jkennedyvz, @mdrxy

View release on GitHub
deepagents-cli==0.0.45 Bug fix

Fixed rejection of out‑of‑tree symlinked AGENTS.md and improved backspace word deletion behavior.

Full changelog

Bug Fixes

  • Reject out-of-tree symlinked project AGENTS.md (#3039) (4618098)
  • Support modified backspace word deletion (#3037) (bc3d019)

Internal maintainers: @emanbareket, @mdrxy

View release on GitHub
deepagents==0.5.5 Bug fix

Harden FilesystemBackend against symlink loops.

Full changelog

Bug Fixes

  • Prevent stdin hang by passing DEVNULL (#2427) (5bf5fae)
  • Harden FilesystemBackend against symlink loops (#3035) (abd02f9)
  • Surface EOF-newline mismatch in edit_file (#3031) (d30686e)

Thanks to our community contributors: @Prithvi1994

Internal maintainers: @eyurtsev, @mdrxy

View release on GitHub
deepagents-code==0.0.1 Maintenance

Minor fixes and improvements.

Changelog

Initial release
chore(code): init (#3027)

Internal maintainers: @mdrxy

View release on GitHub
deepagents-cli==0.0.44 Bug fix

Fixes startup import deadlock

Full changelog

Bug Fixes

Internal maintainers: @mdrxy

View release on GitHub
deepagents-cli==0.0.43 New feature
Notable features
  • Bundled chat frontend for deepagent deploy
Full changelog

Features

  • Bundled chat frontend for deepagent deploy (#2940) (9bde0c7)

Internal maintainers: @victorm-lc

View release on GitHub
deepagents-cli==0.0.42 Bug fix
Notable features
  • `shift+enter` newline support on kitty-capable terminals
  • Enter key behavior hints in `/model` empty state
Full changelog

Features

  • shift+enter newline on kitty-capable terminals (#2869) (34e6614)
  • Hint Enter behavior in /model empty state (#2933) (7cffa16)

Bug Fixes

  • Queue /model switches during server startup (#2895) (6ed7b65)

Thanks to our community contributors: @naveen9871

Internal maintainers: @vishnu-ssuresh, @mdrxy

View release on GitHub
deepagents==0.5.4 New feature
Notable features
  • Harness profiles (ProviderProfile, HarnessProfile) with built-in OpenAI and Anthropic profiles
  • 10-20 point performance lift on tau2-bench with model-specific profiles
  • Third-party plugin support via deepagents.provider_profiles and deepagents.harness_profiles entry points
Full changelog

Highlights

Until this release, deepagents shipped a single set of prompts, tools, and middleware tuned to work across all model families. Today's headline change is harness profiles — a declarative override layer for the parts of the harness that vary per model (system-prompt prefix/suffix, tool inclusion and naming, middleware selection, subagent configuration, skills). ProviderProfile shapes how resolve_model builds the client; HarnessProfile / HarnessProfileConfig shape how create_deep_agent assembles the agent. Both are keyed by provider ("openai") or full provider:model spec ("openai:gpt-5.4"), and registrations are additive — your call site doesn't change when you swap models.

We currently ship built-ins for OpenAI and Anthropic out of the box, drawn directly from each vendor's published prompting guides (Codex's apply_patch / shell_command conventions, Claude's tool-result reflection and active-investigation patterns, etc.). On a curated tau2-bench subset, applying these profiles produced a 10–20 point lift over the default harness (GPT-5.3 Codex: 33% -> 53%; Claude Opus 4.7: 43% -> 53%). Third-party plugins can register via the deepagents.provider_profiles and deepagents.harness_profiles entry points.

See the profiles documentation for the full field surface, merge semantics, and plugin packaging.

Features

Bug Fixes

  • Normalize Windows backslash paths before PurePosixPath processing (#1859) (e1c1d50)
  • Preserve CRLF line endings in sandbox edit (#2899) (291aebe)
  • Support read-your-writes in StateBackend (#2991) (0924869)
  • Treat boundary-truncated UTF-8 in read() prefix check as text (#2980) (c36ebc7)
  • Use configurable directly instead of tracing context for subagent tagging (#2845) (bd6ec6b)

Performance Improvements

  • Add cache breakpoint to MemoryMiddleware (#2713) (1699f3a)

Thanks to our community contributors: @SeongBeomLEE, @pawel-twardziak

Internal maintainers: @sydney-runkle, @vishnu-ssuresh, @DanielKneipp, @hntrl, @kevinbfrank, @ccurme, @jacoblee93, @mdrxy

View release on GitHub
deepagents-cli==0.0.41 Mixed
Notable features
  • Custom auth via [auth] configuration
  • --startup-cmd flag
  • Actionable notifications and update modal
Full changelog

Features

  • cli: --startup-cmd flag (#2841) (8adcc2c)
  • cli: actionable notifications, update modal (#2855) (5fcd368)
  • cli: custom auth via [auth] (#2734) (417ddaa)
  • cli: refresh footer git branch after shell commands (#2851) (ee4fddd)
  • cli: rework /version, add release-age and editable-install guard (#2854) (1ae053f)

Bug Fixes

  • cli: gate agent-swap resume hint on server checkpoint (#2862) (f14b9cd)
  • cli: keep thinking spinner visible during text streaming (#2849) (2cb2244)
  • cli: re-show thinking spinner on hitl/ask_user resume (#2847) (34a6167)
  • cli: restore ctrl+j newline keybinding (#2827) (2e1a3f4)

Thanks to our community contributors: @gojun077, @naveen9871, @NinaadRao (LinkedIn)

Internal maintainers: @mdrxy, @victorm-lc

View release on GitHub
deepagents-cli==0.0.40 New feature
Notable features
  • --max-turns flag for non-interactive mode
  • /agents switcher
Full changelog

Features

Bug Fixes

  • Normalize serialized message dicts in /offload (#2808) (0d11bd4)

Thanks to our community contributors: @ethan6188, @NinaadRao

Internal maintainers: @ramon-langchain, @mdrxy

View release on GitHub
deepagents-acp==0.0.6 Mixed
Notable features
  • Opus 4.7 model support in demo agent
  • Baseten platform integration in demo agent
Full changelog

Features

Bug Fixes

  • Restore passing tests after acp v0.9 schema bump (#2813) (29a351a)

Thanks to our community contributors: @ethan6188

Internal maintainers: @sydney-runkle, @jacoblee93, @mdrxy

View release on GitHub
deepagents-cli==0.0.39 Mixed
Notable features
  • Inline argument hints for slash commands
  • Subagents support for `deepagents deploy` command
Full changelog

Features

  • cli: inline argument hints for slash commands (#2181) (6b58e06)
  • cli: subagents for deepagents deploy (#2786) (7dd5565)

Bug Fixes

  • cli: throttle update notification to once per day and fix teardown banner (#2764) (ba31294)

Thanks to our community contributors: @km322 (LinkedIn)

Internal maintainers: @sydney-runkle, @hari-dhanushkodi, @mdrxy

View release on GitHub
deepagents-cli==0.0.38 New feature
Notable features
  • User scoped memory
  • O(1) message lookups in MessageStore
View release on GitHub
deepagents==0.5.3 Breaking risk
Breaking changes
  • model=None parameter deprecated in create_deep_agent
Notable features
  • Static structured output for subagent responses
View release on GitHub
deepagents-cli==0.0.37 New feature
Notable features
  • Permissions for deepagents deploy
  • Hardened config parsing with unit tests
View release on GitHub
deepagents==0.5.2 New feature
Security fixes
  • Catch PermissionError in FilesystemBackend ripgrep
Notable features
  • Permissions system for filesystem access control
  • upload_files for StateBackend
View release on GitHub
deepagents-cli==0.0.36 New feature
Notable features
  • deepagents deploy command
  • Tavily API key warnings
View release on GitHub
langchain-repl==0.0.1 Feature
Notable features
  • Minimal REPL for testing purposes
Full changelog

Initial release
cr
hotfix(repl): stub package
release(deepagents): 0.5.0 (#2509)
chore(deps): bump the uv group across 1 directory with 3 updates (#2487)
feat(repl): add minimal repl for testing purposes (#2482)

View release on GitHub
langchain-daytona==0.0.5 Mixed
Security fixes
  • Pygments updated to 2.20.0 (GHSA-5239-wwwm-4pmq)
Notable features
  • LangSmith sandbox ported from CLI to SDK
  • Evaluation suite consolidated to standalone libs/evals
View release on GitHub
deepagents-acp==0.0.5 Bug fix
Security fixes
  • Dangerous shell pattern blocking in command auto-approve
Notable features
  • Model switching
View release on GitHub
deepagents-cli==0.0.35 New feature
Notable features
  • Skills invocation via /skill:name and --skill
  • Theme system with color overrides and global dotenv
  • Auto-update lifecycle with /update and /auto-update commands
View release on GitHub
deepagents==0.5.0 Breaking risk
Breaking changes
  • Legacy subagents API removed
Notable features
  • Async subagent middleware for remote LangGraph servers
  • Multimodal support in FileData and backend return types
View release on GitHub
deepagents==0.5.0a4 Breaking risk
Breaking changes
  • Legacy subagents API removed
Security fixes
  • Pygments vulnerability (GHSA-5239-wwwm-4pmq)
  • CVE-2025-68664
Notable features
  • Large HumanMessages eviction
  • Async subagent middleware for remote LangGraph servers
  • FileData multimodal support
View release on GitHub
deepagents==0.5.0a3 Breaking risk
Breaking changes
  • Legacy subagents API removed
Security fixes
  • Pygments vulnerability (GHSA-5239-wwwm-4pmq)
  • CVE-2025-68664
Notable features
  • Large HumanMessages eviction
  • Async subagent middleware
  • FileData multimodal support
View release on GitHub
deepagents==0.5.0a2 Mixed
Security fixes
  • CVE-2025-68664
Notable features
  • async subagent middleware for remote LangGraph servers
  • memory benchmarking framework
  • LangSmith integration metadata
View release on GitHub
deepagents==0.5.0a1 Mixed
Security fixes
  • CVE-2025-68664
Notable features
  • async subagent middleware for remote LangGraph servers
  • memory benchmarking framework
  • LangSmith integration metadata
View release on GitHub
deepagents==0.4.12 Feature
Notable features
  • LangSmithSandbox
Full changelog

Changes since deepagents==0.4.11
release(deepagents): 0.4.12 (#2099)
feat(sdk): port LangSmithSandbox from main to v0.4 (#2098)
chore(sdk,cli): make langchain-anthropic and langchain-google-genai explicit deps (#1886)

View release on GitHub
deepagents-cli==0.0.34 New feature
Notable features
  • External editor support via ctrl+x and /editor
  • Defer HITL approval menu while user typing
Full changelog

Features

  • External editor support via ctrl+x and /editor (#1861) (bf5d088)
  • Defer HITL approval menu while user is typing (#1833) (1d1572e)

Bug Fixes

  • Resolve config-defined providers during runtime model swaps (#1941) (aebc660)

Thanks to our community contributors: @popey, @guptagautam01 (LinkedIn)

View release on GitHub
langchain-quickjs==0.0.1 New feature
Notable features
  • QuickJS REPL with toolruntime support
  • Async support via daemon thread
Full changelog

Initial release
feat(quickjs): support toolruntime (#1932)
chore(quickjs): clean up (#1927)
feat(quickjs): prefer async when on async path (#1922)
feat(quickjs): add async support using a daemon thread (#1918)
chore(sdk): increment version to 0.5 (#1908)
chore(sdk,cli): make langchain-anthropic and langchain-google-genai explicit deps (#1886)
release(sdk): 0.4.11 (#1868)
chore: bump lock (#1871)
feat(quickjs): adding quickjs repl (#1860)

View release on GitHub
deepagents-cli==0.0.33 Breaking risk
Notable features
  • client-server via langgraph dev
  • LiteLLM and Baseten provider integration
  • model status display with performance improvements
View release on GitHub
deepagents==0.4.11 New feature
Notable features
  • LangSmith integration metadata
  • MemoryAgentBench evaluation suite
  • subagent_model parameter
Full changelog

Changes since deepagents==0.4.10
release(sdk): 0.4.11 (#1868)
chore(sdk): make langsmith explicit evals dep (#1867)
chore(sdk): add docs link to LocalShellBackend virtual_mode deprecation warning (#1866)
test(sdk): add ChatBaseten for evals (#1839)
feat(sdk): Add LangSmith integration metadata to deepagents (#1837)
fix(sdk): strip leading slash from glob patterns before matching (#1846)
feat(sdk): add MemoryAgentBench evaluation suite (#1807)
fix(sdk): accept all langsmith tracing env vars in evals conftest (#1847)
test(sdk): add 3 more evals for tool usage (longer trajectories) (#1840)
chore(sdk): add evals README, ensure tracing (#1834)
feat(sdk): add subagent_model param to create_deep_agent (#1369)
refactor(sdk): extract model resolution helpers into _models module (#1825)

View release on GitHub
deepagents-cli==0.0.32 New feature
Notable features
  • token breakdown in /tokens command
  • --json flag for machine-readable output
View release on GitHub
langchain-daytona==0.0.4 New feature
Notable features
  • Polling policy configuration (callable function)
  • Exposed timeout and polling interval parameters
Full changelog

Changes since langchain-daytona==0.0.3
feat(daytona): add polling policy (callable function) (#1797)
release(sdk): prepare 0.4.10 patch release (#1795)
feat(daytona): expose timeout and polling interval params (#1792)
fix(daytona): fix execute implementation (#1756)
release(deepagents): 0.4.9 (#1773)
release(deepagents): 0.4.8 (#1764)

Thanks to our contributors: @sydney-runkle

View release on GitHub
deepagents==0.4.9 Feature
Notable features
  • Prompt guidance for large tool results
Full changelog

Changes since deepagents==0.4.8
release(deepagents): 0.4.9 (#1773)
feat(sdk): add prompt guidance for large tool results (#1763)

Thanks to our contributors: @sydney-runkle

View release on GitHub
deepagents==0.4.8 New feature
Notable features
  • --acp CLI mode for ACP server deployment
  • Factory function for summarization tool middleware
Full changelog

Changes since deepagents==0.4.7
release(deepagents): 0.4.8 (#1764)
test(sdk): bump langsmith-sdk for fixes experiments (#1752)
feat(sdk): add factory function for summarization tool middleware (#1749)
feat(sdk): add solve_rate metric to evals reporter (#1711)
feat(cli): --acp mode to run CLI agent as ACP server (#1297)
test(sdk): add metadata for experiments (#1652)
test(sdk): reuse model fixture in summarization evals (#1697)
feat(sdk): add baseten-hosted models to evals workflow (#1682)

View release on GitHub
deepagents-cli==0.0.31 New feature
Notable features
  • ask_user tool for agent questions
  • thread search, sort, and deletion
  • per-thread working directory tracking
View release on GitHub
deepagents-cli==0.0.30 New feature
Notable features
  • --acp mode for ACP server deployment
  • Model detail footer in CLI
  • Profile override persistence on hot-swap
Full changelog

Features

  • --acp mode to run CLI agent as ACP server (#1297) (c9ba00a)
  • Model detail footer + persist --profile-override on hot-swap (#1700) (f2c8b54)
  • Show message timestamp toast on click (#1702) (4f403ec)

Bug Fixes

  • Expire ctrl+c quit window when toast disappears (#1701) (38b5ea9)
  • Preserve input text when escaping shell/command mode (#1706) (3c00edb)
  • Right-align token count next to model name in status bar (#1705) (311c919)

Thanks to our contributors: @mdrxy

View release on GitHub
deepagents-cli==0.0.29 New feature
Notable features
  • --model-params flag on /model command
  • --shell-allow-list all option
  • Hook dispatch for external tool integration
Full changelog

Features

Bug Fixes

  • Anchor ChatInput below scrollable area (#1671) (11105d9)
    • Remove dead chat-spacer widget and resize handler (#1686) (b6ecec5)

Thanks to our contributors: @jcuypers, @eyurtsev

View release on GitHub
langchain-daytona==0.0.3 Maintenance

Minor fixes and improvements.

Full changelog

Changes since langchain-daytona==0.0.2
release(sdk): bump partner packages (daytona 0.0.3, modal 0.0.2, runloop 0.0.3) (#1689)
release(sdk): bump deepagents to 0.4.7 (#1688)
feat(infra): add fireworks as eval provider (#1684)
release(sdk): bump deepagents to 0.4.6 (#1677)
chore: bump lock file (#1674)
chore(sdk): expand eval matrix and fix model name access for groq (#1623)
release(deepagents): 0.4.5 (#1622)
chore(sdk): expand eval model matrix to 28 models across 7 providers (#1617)
chore: standardize makefile type targets/ add type CI check (#1581)
chore(deps): bump the pip-dependencies group across 8 directories with 17 updates (#1564)
chore(deps): bump the pip-dependencies group across 8 directories with 19 updates (#1491)
release(deepagents): 0.4.4 (#1519)
release(deepagents): 0.4.3 (#1465)
fix(sdk): guard timeout forwarding for older backend packages (#1461)
chore(deepagents): bump min version for langchain-anthropic (#1431)
feat(sdk): add per-command timeout override to execute() (#1154)
chore: auto-generate makefile help from target annotations (#1417)
chore: revert SDK version following yank (#1407)
chore(daytona): add integration test suite to daytona (#1397)
chore: enable type checking for all other packages (#1365)
release(sdk): 0.4.2 (#1312)
chore: rollback uv workspace migration, restore per-package lockfiles (#1278)
chore: migrate to uv workspace (#1264)
chore(deps): bump the pip-dependencies group across 8 directories with 9 updates (#1252)
chore(deps): bump cryptography from 46.0.4 to 46.0.5 in /libs/partners/daytona (#1243)
chore: rename master to main (#984)
release(sdk): 0.4.0 (#1236)
release(deepagents): 0.4 (#1234)
chore(deps): bump the pip-dependencies group across 8 directories with 27 updates (#1230)

Thanks to our contributors: @sydney-runkle, @eyurtsev, @mdrxy

View release on GitHub
langchain-modal==0.0.2 Maintenance

Minor fixes and improvements.

Full changelog

Changes since langchain-modal==0.0.1
release(sdk): bump partner packages (daytona 0.0.3, modal 0.0.2, runloop 0.0.3) (#1689)
release(sdk): bump deepagents to 0.4.7 (#1688)
feat(infra): add fireworks as eval provider (#1684)
release(sdk): bump deepagents to 0.4.6 (#1677)
chore: bump lock file (#1674)
chore(sdk): expand eval matrix and fix model name access for groq (#1623)
release(deepagents): 0.4.5 (#1622)
chore(sdk): expand eval model matrix to 28 models across 7 providers (#1617)
chore: standardize makefile type targets/ add type CI check (#1581)
test: add standard tests for runloop and modal (#1537)
release(deepagents): 0.4.4 (#1519)
release(deepagents): 0.4.3 (#1465)
fix(sdk): guard timeout forwarding for older backend packages (#1461)
chore(deepagents): bump min version for langchain-anthropic (#1431)
feat(sdk): add per-command timeout override to execute() (#1154)
chore: auto-generate makefile help from target annotations (#1417)
chore: revert SDK version following yank (#1407)
chore: enable type checking for all other packages (#1365)
release(sdk): 0.4.2 (#1312)
chore: rollback uv workspace migration, restore per-package lockfiles (#1278)
chore: migrate to uv workspace (#1264)
chore(deps): bump the uv group across 2 directories with 1 update (#1245)
chore: rename master to main (#984)
release(sdk): 0.4.0 (#1236)
release(deepagents): 0.4 (#1234)

Thanks to our contributors: @sydney-runkle, @eyurtsev, @mdrxy

View release on GitHub
langchain-runloop==0.0.3 Maintenance

Minor fixes and improvements.

Full changelog

Changes since langchain-runloop==0.0.2
release(sdk): bump partner packages (daytona 0.0.3, modal 0.0.2, runloop 0.0.3) (#1689)
release(sdk): bump deepagents to 0.4.7 (#1688)
feat(infra): add fireworks as eval provider (#1684)
release(sdk): bump deepagents to 0.4.6 (#1677)
chore: bump lock file (#1674)
chore(sdk): expand eval matrix and fix model name access for groq (#1623)
release(deepagents): 0.4.5 (#1622)
chore(sdk): expand eval model matrix to 28 models across 7 providers (#1617)
chore: standardize makefile type targets/ add type CI check (#1581)
test: add standard tests for runloop and modal (#1537)
release(deepagents): 0.4.4 (#1519)
release(deepagents): 0.4.3 (#1465)
fix(sdk): guard timeout forwarding for older backend packages (#1461)
chore(deepagents): bump min version for langchain-anthropic (#1431)
feat(sdk): add per-command timeout override to execute() (#1154)
chore: auto-generate makefile help from target annotations (#1417)
chore: revert SDK version following yank (#1407)
chore: enable type checking for all other packages (#1365)
release(sdk): 0.4.2 (#1312)
chore: rollback uv workspace migration, restore per-package lockfiles (#1278)
chore: migrate to uv workspace (#1264)
chore(deps): bump the uv group across 2 directories with 1 update (#1245)
chore: rename master to main (#984)
release(sdk): 0.4.0 (#1236)
release(deepagents): 0.4 (#1234)

Thanks to our contributors: @sydney-runkle, @eyurtsev, @mdrxy

View release on GitHub
deepagents==0.4.7 Bug fix

Restores full routed paths in file write/edit results.

Full changelog

Changes since deepagents==0.4.6
release(sdk): bump deepagents to 0.4.7 (#1688)
fix(deepagents): restore full routed path in write/edit results (#1681)
feat(infra): add fireworks as eval provider (#1684)

Thanks to our contributors: @naga-k

View release on GitHub
deepagents==0.4.6 Bug fix

Path boundary enforcement and large output text extraction fixes.

View release on GitHub
deepagents-cli==0.0.28 New feature
Notable features
  • Video support for multimodal inputs
  • NVIDIA API key support with default model
  • Fuzzy search for slash command autocomplete
Full changelog

Features

Bug Fixes

  • Backspace at cursor position 0 exits mode even with text (#1666) (dfa4c1f)
  • Skip auto-approve toggle when modal screen is open (#1668) (6597f0b)
  • Truncate model name in status bar on narrow terminals (#1665) (0e24a04)

Thanks to our contributors: @arthur-drozdov, @soumilinandi, @mdrxy

View release on GitHub
deepagents-cli==0.0.27 New feature
Notable features
  • fuzzy model search in switcher
  • model usage statistics display
  • substring command history matching
View release on GitHub
deepagents-cli==0.0.26 New feature
Notable features
  • /compact command
  • model profile overrides
  • summarization status display
View release on GitHub
deepagents==0.4.5 New feature
Notable features
  • /compact command for response compaction
  • Expanded eval matrix to 28 models across 7 providers
Full changelog

Changes since deepagents==0.4.4
release(deepagents): 0.4.5 (#1622)
chore(sdk): expand eval model matrix to 28 models across 7 providers (#1617)
chore: remove unnecessary @pytest.mark.asyncio decorators (#1616)
chore(sdk): update evals to use chat model instance (#1614)
fix(sdk): make evals more robust across model output variations (#1608)
fix(infra): streamline evals workflow and summary (#1602)
docs(sdk): improve docstrings around summarization conditions (#1597)
chore(sdk): add json report to evals (#1594)
chore(sdk,cli): switch claude-sonnet-4-5-20250929 -> claude-sonnet-4-6 (#1590)
feat(cli): /compact command (#1579)
chore: standardize makefile type targets/ add type CI check (#1581)
ci(sdk): skip some evals (#1580)
docs(sdk): rename 'prompt' to 'system_prompt' in subagent keys (#1578)
docs(sdk): docstring fixes for summarization.py (#1568)
chore(sdk): remove unused base_path alias in skills.py (#1543)
chore(deps): bump the pip-dependencies group across 8 directories with 17 updates (#1564)
fix(sdk,cli,deps): support timeout=0 for sandbox execute() (#1558)
fix: Potential fixes for 2 code quality findings (#1534)
fix: Variable defined multiple times (#1536)
feat(cli,sdk): compaction hook (#1420)
chore(deps): bump the pip-dependencies group across 8 directories with 19 updates (#1491)

Thanks to our contributors: @AdemBoukhris457, @AlexPapadakis, @eyurtsev

View release on GitHub
deepagents==0.4.4 Bug fix

Path handling fixes and memory loading performance optimization.

View release on GitHub
deepagents-cli==0.0.25 Breaking risk
Breaking changes
  • Environment variable fallback for model configuration removed
Notable features
  • OpenRouter header configuration
  • Default model changed to gemini-3.1-pro-preview
Full changelog

Features

  • Set OpenRouter headers, default to gemini-3.1-pro-preview (#1455) (95c0b71), closes #1454

Bug Fixes

View release on GitHub
deepagents==0.4.3 Security relevant
Security fixes
  • Directory traversal via unsanitized glob pattern in glob_info()
  • Path traversal via Glob pattern validation bypass
Notable features
  • Evaluation tests for skills and memory
Full changelog

Changes since deepagents==0.4.2
release(deepagents): 0.4.3 (#1465)
fix(sdk): guard timeout forwarding for older backend packages (#1461)
fix(sdk): correct imports and mock inheritance in filesystem middleware tests (#1464)
fix(sdk): warn on virtual mode (#1462)
chore(sdk): add more eval test cases (#1452)
test(deepagents): add evaluation tests for skills (#1429)
fix(deepagents): Fix Memory prompt guidelines that could lead to wrong actions (#1448)
test(sdk): add smoke test for system prompt with memory and skills (#1430)
chore(deepagents): bump min version for langchain-anthropic (#1431)
feat(deepagents): add evaluation tests for agent memory (#1415)
chore(sdk): adapt integration tests to evals framework (#1426)
feat(sdk): add per-command timeout override to execute() (#1154)
chore: auto-generate makefile help from target annotations (#1417)
chore(sdk): add more evals (#1404)
chore(sdk): minor updates to system message (#1367)
chore(sdk): add model parameterization to evals (#1408)
chore: revert SDK version following yank (#1407)
chore(sdk): add simple eval scaffolding (#1400)
refactor(sdk): add logging to silent path skips, rename validate_path (#1370)
fix(cli,sdk): harden path hardening (#918)
chore(sdk): add smoke tests for system message (#1363)
feat(sdk): enable type checking in deepagents and resolve most linting issues (#991)
chore(sdk): move some integration tests into unit tests (#1330)
revert(sdk): reimplement skill loading via skill tool calling (#1328)
fix(sdk): missing _validate_path() on Glob tool's path parameter allows directory traversal (#1320)
fix: Path Traversal via Unsanitized Glob Pattern in glob_info() Bypasses Virtual Mode Sandbox (#1322)
feat(sdk): add base agent prompt (#1314)

Thanks to our contributors: @MuriloAndre2000, @vtrivedy, @sydney-runkle, @app/corridor-security, @app/copilot-swe-agent, @ccurme, @eyurtsev, @mdrxy

View release on GitHub
deepagents-cli==0.0.24 New feature
Notable features
  • single-click Rich hyperlink support
  • windowed thread hydration with limits
  • /tokens command for context window
View release on GitHub
deepagents-cli==0.0.23 New feature
Notable features
  • drag-and-drop image attachment
  • skill deletion command
  • visual mode indicators
View release on GitHub
deepagents-cli==0.0.22 New feature
Notable features
  • OpenRouter provider support
  • system and default prompt updates
View release on GitHub
deepagents-acp==0.0.4 Security relevant
Security fixes
  • Command type extraction ignores piped commands, enabling privilege escalation
Full changelog

Changes since deepagents-acp==0.0.3
chore(acp): release 0.0.4 (#1325)
fix: ACP Command Type Extraction Ignores Piped Commands, Enabling Privilege Escalation via "Always Allow" (#1321)

Thanks to our contributors: @app/corridor-security

View release on GitHub
deepagents==0.4.2 New feature

Reimplement skill loading via skill tool calling.

Full changelog

Changes since deepagents==0.4.1
release(sdk): 0.4.2 (#1312)
feat(sdk): reimplement skill loading via skill tool calling (#1306)

Thanks to our contributors: @sydney-runkle

View release on GitHub
deepagents-acp==0.0.3 Feature
Notable features
  • Execute tool handling
  • Allow-type permissions
  • Demo coding agent
Full changelog

Changes since deepagents-acp==0.0.2
release(acp): 0.0.3 (#1310)
feat(acp): Adds execute tool handling, allow of type permissions, demo coding agent (#1302)
chore(acp): automatically attach checkpointer if missing (#1294)

Thanks to our contributors: @jacoblee93

View release on GitHub
deepagents-cli==0.0.21 New feature
Notable features
  • /threads command switcher with clickable links
  • /trace command for LangSmith
  • /changelog, /feedback, /docs commands
View release on GitHub
deepagents==0.4.1 New feature
Notable features
  • Native image content blocks from read_file
Full changelog

Changes since deepagents==0.4.0
feat(sdk): return native image content blocks from read_file (#1163)
chore(deps): bump the pip-dependencies group across 8 directories with 9 updates (#1252)
refactor(sdk): use public SummarizationMiddleware middleware (#1241)
chore: rename master to main (#984)

Thanks to our contributors: @app/dependabot, @mdrxy

View release on GitHub
deepagents-cli==0.0.20 New feature
Notable features
  • Model switcher and arbitrary chat model support
  • Non-interactive mode with shell allow-listing
  • Built-in skills with skill-creator
View release on GitHub
deepagents==0.4.0 New feature
Notable features
  • langchain-modal, langchain-daytona, langchain-runloop integration packages
  • Refactored summarization middleware with accurate token counting
  • Context overflow exception handling for ChatOpenAI and ChatAnthropic
Full changelog

Changes since deepagents==0.3.12

New integration packages for pluggable sandboxes: langchain-modal, langchain-daytona, and langchain-runloop. See sandboxes guide and example data analysis tutorial.

Notable changes from #1234

  • Refactor the SDK's summarization middleware.
    • Summarization now happens in the model node via wrap_model_call events. Due to this we retain the full message history in the graph state.
    • System messages and tool definitions are accurately counted toward summarization triggers.
    • Add exception handling for context overflows: ChatOpenAI and ChatAnthropic now raise a standard LangChain exception when input exceeds a model's token limit. We catch this error and trigger summarization.
  • When model strings in create_deep_agent are prefixed with 'openai:', we default to the Responses API.

chore(sdk): add support for , in allowed-tools, fail gently for bullet list (#1235)
fix(sdk): relax skills parsing for allowed tools (#1232)
chore(sdk)!: move sandbox provider back to cli (#1226)
refactor(sdk): move function-level test imports out to top (#1168)
fix(sdk): harden skills metadata parsing, improve spec compliance (#1189)
style: update brand names (#1180)

Thanks to our contributors: @sydney-runkle, @eyurtsev, @mdrxy

View release on GitHub
langchain-runloop==0.0.2 Maintenance

Changes since langchain-runloop==0.0.1 release: runloop 0.0.2 (#1219)

View release on GitHub
langchain-daytona==0.0.2 New feature
Notable features
  • Daytona sandbox provider integration package
Full changelog

Initial release
release(daytona): 0.0.2 (#1213)
chore: standardize init in partner packages (#1211)
chore(daytona): minor fix (#1210)
feat: add partner packages for modal and runloop (#1207)
chore(deps): bump the pip-dependencies group across 3 directories with 5 updates (#1183)
release(sdk): deepagents 0.3.12 (#1175)
revert(deepagents): refactor summarization middleware (#1172)
fix(deepagents): refactor summarization middleware (#1138)
chore(daytona): refactor name (#1149)
chore(daytona): add integration package for daytona (#1140)

Thanks to our contributors: @app/dependabot, @eyurtsev

View release on GitHub
langchain-runloop==0.0.1 Maintenance

Initial release chore: standardize init in partner packages (#1211) chore(infra): add modal and runloop to ci (#1208) feat: add partner packages for modal and runloop (#1207)

View release on GitHub
langchain-modal==0.0.1 Maintenance

Initial release chore: standardize init in partner packages (#1211) chore(infra): add modal and runloop to ci (#1208) feat: add partner packages for modal and runloop (#1207)

View release on GitHub
deepagents==0.3.12 Maintenance

Minor fixes and improvements.

Full changelog

Changes since deepagents==0.3.11

  • feat(sdk): refactor namespace gen for StoreBackend (#1169)
  • fix(sdk): add explicit NotImplementedError to protocol method bodies (#1152)
  • refactor(cli): reorder property -> method order in SandboxBackendProtocol (#1153)
  • feat(sdk): add sandbox provider exceptions (#1108)

Thanks to our contributors: @eyurtsev, @mdrxy

View release on GitHub
deepagents-cli==0.0.19 New feature
Notable features
  • LocalShellBackend for subagent shell access
  • CJK file mention parsing
  • click and hover autocomplete styling
View release on GitHub
deepagents-acp==0.0.1 New feature
Notable features
  • ACP implementation
Full changelog

Initial release
feat(acp): add acp implementation (#865)
chore: correct brand name (#1025)
chore: nit versions (#997)
chore: enrich pyproject.toml files (#996)
chore(deps): bump the uv group across 3 directories with 6 updates (#905)
fix(sdk,acp,harbor): don't allow relative imports (#882)
chore: update Makefile targets, add pre-commit hooks (#790)
chore(deps): bump the uv group across 3 directories with 3 updates (#796)
fix: import rules (#763)
chore: automatically sort imports (#740)
chore(acp): clean up tests (#604)
feat(acp): draft acp support (#589)

Thanks to our contributors: @app/dependabot, @eyurtsev, @mdrxy

View release on GitHub
langchain_daytona==0.0.1 Maintenance

Initial release chore(daytona): add integration package for daytona (#1140) --- Thanks to our contributors: @eyurtsev

View release on GitHub
deepagents-cli==0.0.18 Bug fix
Notable features
  • Support .agents/skills directory alias
  • Resume thread enhancements
  • Help flag support
Full changelog

Features

Bug Fixes

Thanks to our community contributors: @kaushikkumarkr

View release on GitHub
deepagents==0.3.11 New feature
Notable features
  • Subagent skills support
  • Middleware refactor
Full changelog

Changes since deepagents==0.3.10

  • feat(sdk): subagent skills, middleware refactor (#1041)
  • fix(sdk): add newlines in tool result message for large outputs (#1053)

Thanks to our contributors: @sydney-runkle

View release on GitHub
deepagents-cli==0.0.17 New feature
Notable features
  • Expandable shell command display in HITL approval
  • Model identity support
  • Sandbox provider interface
Full changelog

Features

  • cli: add expandable shell command display in HITL approval (#976)
  • cli: model identity (#770)
  • cli: sandbox provider interface (#900)

Thanks to our contributors: @sydney-runkle, @jkennedyvz, @eyurtsev, @mdrxy

View release on GitHub
deepagents==0.3.10 New feature
Notable features
  • Sandbox provider interface
  • LocalShellBackend implementation
  • Literal grep search, improved file path handling
Full changelog

Changes since deepagents==0.3.9

  • fix(sdk): grep should perform literal search instead of regex (#975)
  • feat(sdk): sandbox provider interface (#900)
  • fix(sdk): handle exact file paths in grep and glob operations (#1017)
  • fix: add py.typed to deepagents (#1024)
  • fix(sdk): return error message instead of raising ValueError for invalid paths (#994)
  • fix(deepagents): prevent state merge error in parallel sub-agents (#954)
  • chore: enrich pyproject.toml files (#996)
  • chore(sdk,cli): add VERSION (#983)
  • feat(sdk): add LocalShellBackend (#930)
  • fix(sdk): string interpolation in BaseSandbox write/edit tools (#955)

Thanks to our contributors: @vtrivedy, @sydney-runkle, @sandeepyadav1478, @Yo-sure, @trindadetiago, @eyurtsev, @mdrxy

View release on GitHub
deepagents-cli==0.0.16 New feature
Notable features
  • configurable ShellMiddleware timeout
  • per-command timeout override
  • thread ID display at splash
View release on GitHub
deepagents-cli==0.0.15 Maintenance

Changes since deepagents-cli==0.0.14 release(cli): 0.0.15 (#953) feat(cli): custom subagents (#916)

View release on GitHub
deepagents-cli==0.0.14 New feature
Notable features
  • VertexAI provider support
  • Model capability validation
Full changelog

Changes since deepagents-cli==0.0.13

release(cli): 0.0.14 (#952)
fix(cli): tool rendering in TUI (#945)
release(deepagents): 0.3.9 (#944)
fix(sdk): summarization offloading (#920)
docs: update cli cover photo (#929)
fix(cli): scrolling wrapping on input/resume, running tool spinner (#913)
feat(cli): Support VertexAI provider on the CLI (#828)
chore(sdk): ReadMe cleanup (#906)
feat: lazily adding trace link to deepagents banner (#894)
fix: default to 5.2 (#896)
feat(sdk): model capability validation (#733)
chore: update Makefile targets, add pre-commit hooks (#790)
chore(cli): update outdated tests (#881)

View release on GitHub
deepagents==0.3.9 Bug fix
Notable features
  • Middle truncation for large content instead of end
  • Tool message truncation from read_file
Full changelog

Changes since deepagents==0.3.8

release(deepagents): 0.3.9 (#944)
fix(sdk): summarization offloading (#920)
refactor: filesystem middleware (#928)
chore(deps): bump the uv group across 3 directories with 6 updates (#905)
test(sdk): add unit test for handling large single-line file bug (#813)
feat(deepagents): truncate middle of content instead of end on tool response eviction (#901)
feat(deepagents): truncate tool messages from read_file (#897)
fix(sdk,acp,harbor): don't allow relative imports (#882)
fix(sdk): handle large payloads and headless container execution (#872)
chore: update Makefile targets, add pre-commit hooks (#790)

View release on GitHub
deepagents-cli==0.0.13 Mixed
Notable features
  • Textual-based terminal UI refactor
  • SDK memory integration
  • Local context middleware for system prompts
Full changelog

Changes since deepagents-cli==0.0.12

release(sdk): bump version (#879)
fix(sdk): make sure that tool truncation applies to execute (#547)
fix(cli): hitl spinner errors (#861)
fix(cli): Improve HITL approval UX (#859)
release: patch release 0.3.7 (#869)
fix(cli): resume should load previous thread messages (#862)
chore(cli): clean checks in cli spacing (#854)
fix(cli): avoid jumping (#853)
chore(cli): turn on more linting (#852)
Add current model display to status bar (#844)
Disable double message submission while agent is working
update remember prompt to work with memory and skills (#842)
feat(cli): focus input when clicking anywhere in the terminal (#826)
feat: summarization offloading (#742)
Bump version to 0.3.7a1 (#817)
chore(deps): bump the uv group across 5 directories with 1 update (#811)
fix(infra): exclude build/ from typechecking (#808)
fix(sdk): BaseSandbox.ls_info() to return absolute paths (#797)
chore: bump deepagents-cli to 0.0.13a2 (#795)
docs: add testing readme (#788)
fix(cli): include tcss and py.typed in package data (#781)
feat(cli): format file tree with markdown (#782)
fix(cli): add explicit package inclusion for setuptools (#780)
add prompt seeding with -m flag (#755)
docs: update model configuration details in README (#772)
fix: import rules (#763)
release(deepagents-cli): 0.0.13a1 (#756)
cli-token-tracking-fixes (#706)
release: deepagents 0.3.6 (#752)
chore: automatically sort imports (#740)
Add LangSmith tracing status to welcome banner (#741)
feat(cli): inject local context into system prompt via LocalContextMiddleware
fix: don't allow Rich markup from user content (#704)
fix(cli): remove duplicate version from welcome.py (#737)
feat(cli): add --version//version commands (#698)
minor release(deepagents): bump version to 0.3.5 (#695)
Port SDK Memory to CLI (#691)
fix thread id (#692)
chore(ci): add uv lock checks (#681)
update version bounds (#687)
CLI Refactor to Textual (#686)
Fix invalid YAML in skill-creator SKILL.md frontmatter (#675)
feat(deepagents): add skills to sdk (#591)
docs: replace gemini 1.5 (#653)
feat(cli): show version in splash screen (#610)
chore(cli): expose version (#609)
fix(cli): handle read_file offset exceeding file length by returning all lines (issue #559) (#568)
chore(cli): remove line (#601)

View release on GitHub
deepagents==0.3.7 New feature
Notable features
  • Summarization prompt improvements
  • Truncation of old write/edit operations in message history
Full changelog

Changes since deepagents==0.3.6

fix(sdk): don't dedent summarization prompt (#870)
feat(deepagents): truncate old write / edit calls in message history (#806)
release: patch release 0.3.7 (#869)
chore(deepagents): add end to end tests to confirm file reducer working properly in state backend (#754)
chore: improve filesystem and subagents tool descriptions (#807)
docs(SDK): clarify usage of file system backend (#850)
nit: standardize naming (#849)
docs(sdk): FilesystemBackend refs fixes (#791)
feat: show allowed tools for each skill in skill list (#837)
fix(sdk): FilesystemMiddleware forward 'name' attribute in large_tool_result from the original tool msg (#825)
docs(sdk): docstring formatting nits (#824)
feat: summarization offloading (#742)
Bump version to 0.3.7a1 (#817)
Add Async ops to Store Backend (#816)
chore(deepagents): add tests for grep in end to end tests (#805)
chore(deepagents): bump langchain in lock file (#800)
chore(deps): bump the uv group across 5 directories with 1 update (#811)
fix(deepagents): respect continuation markers when reading files (#809)
fix(infra): exclude build/ from typechecking (#808)
feat: support SystemMessage for parity w/ create_agent (#803)
chore(deepagnets): end to end tests for agent writing editing files (#804)
fix(sdk): BaseSandbox.ls_info() to return absolute paths (#797)
fix(deepagents): truncate lines on read (#784)
chore(deps): bump the uv group across 3 directories with 3 updates (#796)
fix: refinements for test_summarization (#786)
docs: fix old URLs (#787)
docs: add testing readme (#788)
fix: added error catching for file operations without permissions (#734)
docs(deepagents): update subagent spec (#785)
chore(deepagents): add mini eval for summarization (#751)
docs(sdk): improve FileSystemBackend ref docs (#783)

View release on GitHub
deepagents==0.3.7a1 Security relevant
Security fixes
  • Fix CVE-2025-68664
Notable features
  • Async operations in Store Backend
  • Improved file operations and error handling
View release on GitHub
deepagents-cli==0.0.13a2 New feature
Notable features
  • Local context injection via LocalContextMiddleware
  • Skill system integration
  • Token tracking and counting
Full changelog

Initial release

chore: bump deepagents-cli to 0.0.13a2 (#795)
docs: add testing readme (#788)
fix(cli): include tcss and py.typed in package data (#781)
feat(cli): format file tree with markdown (#782)
fix(cli): add explicit package inclusion for setuptools (#780)
add prompt seeding with -m flag (#755)
docs: update model configuration details in README (#772)
fix: import rules (#763)
release(deepagents-cli): 0.0.13a1 (#756)
cli-token-tracking-fixes (#706)
release: deepagents 0.3.6 (#752)
chore: automatically sort imports (#740)
Add LangSmith tracing status to welcome banner (#741)
feat(cli): inject local context into system prompt via LocalContextMiddleware
fix: don't allow Rich markup from user content (#704)
fix(cli): remove duplicate version from welcome.py (#737)
feat(cli): add --version//version commands (#698)
minor release(deepagents): bump version to 0.3.5 (#695)
Port SDK Memory to CLI (#691)
fix thread id (#692)
chore(ci): add uv lock checks (#681)
update version bounds (#687)
CLI Refactor to Textual (#686)
Fix invalid YAML in skill-creator SKILL.md frontmatter (#675)
feat(deepagents): add skills to sdk (#591)
docs: replace gemini 1.5 (#653)
feat(cli): show version in splash screen (#610)
chore(cli): expose version (#609)
fix(cli): handle read_file offset exceeding file length by returning all lines (issue #559) (#568)
chore(cli): remove line (#601)
minor version bump, model setting, agent skill spec support, skill creator example (#600)
Comply with Anthropic Agent Skills spec (#592)
feat(cli): add --model flag with auto-detection (#584)
feat: add skill-creator skill with init and validation scripts (#579)
docs(cli): add LangSmith environment variables documentation (#583)
CLI release (#581)
feat(cli): add DEEPAGENTS_LANGSMITH_PROJECT configuration (#577)
feat: add ability to paste images in input (#555)
chore(harbor/cli): allow benchmarking with either cli or SDK (#542)
chore(cli): add comprehensive testing for sandbox operations (#501)
test make format (#483)
Update README to specify CLI (#490)
docs(cli): enhance README with comprehensive documentation (#489)
fix(cli): for now use non peristent implementation of shell (#488)
chore(cli): bump lock file (#487)
chore(cli): add end to end test to the cli (#482)
release(deepagents, cli) (#477)
Harrison/fix diffing (#478)
truncate glob (#473)
fix(cli): 2nd argument must be called runtime (not _runtime) (#472)
add file upload and download apis (#381)
chore(cli): other lints (#464)
feat: add option to disable splash on startup (#446)
chore(cli): pull out interrupt on config (#463)
Harrison/add gemini support (#456)
chore(cli): remove internal file that's not needed (#462)
chore(cli): apply auto-fixes for linting (#461)
chore(cli): quick linting fixes (#460)
chore(cli): remove hard-coded paths (#458)
cli: inherit env variables for cli (#459)
fix(deepagents-cli): fix linting (broke CI) (#457)
feat(cli): add project-level skills support (#439)
fix: localize key bindings and update tips for macOS compatibility (#451)
chore: cleanup markdownlint errors in README.md (#447)
fix cli rendering (#445)
add auto accept option to CLI (#421)
Remove unnecessary dependencies from deepagents module (#429)
fix: use request.override instead of direct attribute overrides (#431)
add missing type annotations (#425)
chore(deepagent-cli): remove double diff display (WIP) (#387)
Add skills and dual-scope memory to deepagents CLI (#315)
use thread id rather than hardcoding to main (#423)
release: deepagents 0.2.9, cli 0.0.9 (#411)
patch: remove resumable shell middleware (#410)
chore(cli): internal refactor and some unit tests for tool descriptions (#394)
Add simple benchmark tests (#395)
fix: remove temperature, not supported by some OpenAI models (o3) (#392)
release 0.0.8 deepagents-cli (#390)
release (#388)
clean up for HIL logic (#384)
chore: finish migration into monorepo structure (#383)
chore: clean up placeholder test file (#378)
feat(sandbox-protocol): introduce id property + restore missing traceback (#379)
chore: add simple integration tests (#377)
chore: carve out integration tests for CLI (#376)
chore: clean up some sandbox provider details (#375)
feat: sandbox protocol (#319)
chore: quick linting pass in cli (#349)
fix(cli): handle multiple concurrent interrupts in HITL workflow (#318)
feat(cli): add fetch_url tool for converting web content to markdown (#310)
ctrl-c protection, add buffer to avoid accidentally exiting thread (#300)
update message.text() to .text (#317)
release 0.2.5 (#306)
fix-cli(ui): show tool call running in spinner instead of hanging cursor (#305)
cli-update: make execute_task async, allows abort to work (#299)
fix (cli-ui): remove brittle/dead code for summarization tracking (#298)
fix (cli-ux): improved diff viewer (highlighting, line_nums, wrapping), fixed spacing nits on tool approval (#293)
ci: enable format lint (#292)
fix (cli): autocomplete for @ and / commands through directories, bash mode in TUI (#278)
chore: delete accidental file (#282)
handle hilt in a way that makes tracing better (#277)
simplify hilt (#276)
chore: add ci (#254)
move agent memory and bump (#249)
Revert "fix(cli): Abort functionality, autocomplete, and fixed memory prompts…" (#248)
fix(cli): Abort functionality, autocomplete, and fixed memory prompts (#246)
cr (#245)
fix(deepagents-cli): package the agent prompt correctly (#242)
release
ix(cli): Fix token counting bugs and improve clarity (#240)
release (#238)
Add deepagents cli scaffolding (#224)

View release on GitHub
deepagents-cli==0.0.13a1 New feature
Notable features
  • Textual TUI
  • Skills support
  • Memory integration
Full changelog

Initial release

release(deepagents-cli): 0.0.13a1 (#756)
cli-token-tracking-fixes (#706)
release: deepagents 0.3.6 (#752)
chore: automatically sort imports (#740)
Add LangSmith tracing status to welcome banner (#741)
feat(cli): inject local context into system prompt via LocalContextMiddleware
fix: don't allow Rich markup from user content (#704)
fix(cli): remove duplicate version from welcome.py (#737)
feat(cli): add --version//version commands (#698)
minor release(deepagents): bump version to 0.3.5 (#695)
Port SDK Memory to CLI (#691)
fix thread id (#692)
chore(ci): add uv lock checks (#681)
update version bounds (#687)
CLI Refactor to Textual (#686)
Fix invalid YAML in skill-creator SKILL.md frontmatter (#675)
feat(deepagents): add skills to sdk (#591)
docs: replace gemini 1.5 (#653)
feat(cli): show version in splash screen (#610)
chore(cli): expose version (#609)
fix(cli): handle read_file offset exceeding file length by returning all lines (issue #559) (#568)
chore(cli): remove line (#601)
minor version bump, model setting, agent skill spec support, skill creator example (#600)
Comply with Anthropic Agent Skills spec (#592)
feat(cli): add --model flag with auto-detection (#584)
feat: add skill-creator skill with init and validation scripts (#579)
docs(cli): add LangSmith environment variables documentation (#583)
CLI release (#581)
feat(cli): add DEEPAGENTS_LANGSMITH_PROJECT configuration (#577)
feat: add ability to paste images in input (#555)
chore(harbor/cli): allow benchmarking with either cli or SDK (#542)
chore(cli): add comprehensive testing for sandbox operations (#501)
test make format (#483)
Update README to specify CLI (#490)
docs(cli): enhance README with comprehensive documentation (#489)
fix(cli): for now use non peristent implementation of shell (#488)
chore(cli): bump lock file (#487)
chore(cli): add end to end test to the cli (#482)
release(deepagents, cli) (#477)
Harrison/fix diffing (#478)
truncate glob (#473)
fix(cli): 2nd argument must be called runtime (not _runtime) (#472)
add file upload and download apis (#381)
chore(cli): other lints (#464)
feat: add option to disable splash on startup (#446)
chore(cli): pull out interrupt on config (#463)
Harrison/add gemini support (#456)
chore(cli): remove internal file that's not needed (#462)
chore(cli): apply auto-fixes for linting (#461)
chore(cli): quick linting fixes (#460)
chore(cli): remove hard-coded paths (#458)
cli: inherit env variables for cli (#459)
fix(deepagents-cli): fix linting (broke CI) (#457)
feat(cli): add project-level skills support (#439)
fix: localize key bindings and update tips for macOS compatibility (#451)
chore: cleanup markdownlint errors in README.md (#447)
fix cli rendering (#445)
add auto accept option to CLI (#421)
Remove unnecessary dependencies from deepagents module (#429)
fix: use request.override instead of direct attribute overrides (#431)
add missing type annotations (#425)
chore(deepagent-cli): remove double diff display (WIP) (#387)
Add skills and dual-scope memory to deepagents CLI (#315)
use thread id rather than hardcoding to main (#423)
release: deepagents 0.2.9, cli 0.0.9 (#411)
patch: remove resumable shell middleware (#410)
chore(cli): internal refactor and some unit tests for tool descriptions (#394)
Add simple benchmark tests (#395)
fix: remove temperature, not supported by some OpenAI models (o3) (#392)
release 0.0.8 deepagents-cli (#390)
release (#388)
clean up for HIL logic (#384)
chore: finish migration into monorepo structure (#383)
chore: clean up placeholder test file (#378)
feat(sandbox-protocol): introduce id property + restore missing traceback (#379)
chore: add simple integration tests (#377)
chore: carve out integration tests for CLI (#376)
chore: clean up some sandbox provider details (#375)
feat: sandbox protocol (#319)
chore: quick linting pass in cli (#349)
fix(cli): handle multiple concurrent interrupts in HITL workflow (#318)
feat(cli): add fetch_url tool for converting web content to markdown (#310)
ctrl-c protection, add buffer to avoid accidentally exiting thread (#300)
update message.text() to .text (#317)
release 0.2.5 (#306)
fix-cli(ui): show tool call running in spinner instead of hanging cursor (#305)
cli-update: make execute_task async, allows abort to work (#299)
fix (cli-ui): remove brittle/dead code for summarization tracking (#298)
fix (cli-ux): improved diff viewer (highlighting, line_nums, wrapping), fixed spacing nits on tool approval (#293)
ci: enable format lint (#292)
fix (cli): autocomplete for @ and / commands through directories, bash mode in TUI (#278)
chore: delete accidental file (#282)
handle hilt in a way that makes tracing better (#277)
simplify hilt (#276)
chore: add ci (#254)
move agent memory and bump (#249)
Revert "fix(cli): Abort functionality, autocomplete, and fixed memory prompts…" (#248)
fix(cli): Abort functionality, autocomplete, and fixed memory prompts (#246)
cr (#245)
fix(deepagents-cli): package the agent prompt correctly (#242)
release
ix(cli): Fix token counting bugs and improve clarity (#240)
release (#238)
Add deepagents cli scaffolding (#224)

View release on GitHub
deepagents==0.3.6 Bug fix

Fix subagent exception handling and add agent naming.

Full changelog

Changes since deepagents==0.3.5

chore: add tests for config, context, metadata propagation (#776)
fix: import rules (#763)
fix(deepagents): throw a clear exception message when 'messages' key missing from the output of the subagent (#678)
release: deepagents 0.3.6 (#752)
chore: automatically sort imports (#740)
fix: no f string needed (#750)
feat: add agent name when creating subagents (#735)
feat(cli): inject local context into system prompt via LocalContextMiddleware
chore: update twitter URL (#727)
chore: CR on docstrings/readme (#722)

View release on GitHub
deepagents==0.3.5 Bug fix

Minor fixes and improvements.

Full changelog

Changes since deepagents==0.3.4

minor release(deepagents): bump version to 0.3.5 (#695)
Port SDK Memory to CLI (#691)
chore(deepagents): turn on strict linting and add ratchet (#690)
chore(deepagents): add tests for todos middleware (#667)
fix(deepagents): fix content block handling in filesystem middleware tool result eviction (#674)

View release on GitHub
deepagents==0.3.4 Maintenance

Minor fixes and improvements.

Full changelog

Changes since deepagents==0.3.3

Minor release memory bump (#689)
feat: update memory prompt to be more barebones (#685)
improvements to README, reference docs (#688)
update version bounds (#687)

View release on GitHub
deepagents==0.3.3 Bugfix

Changes since deepagents==0.3.2 Deepagents release minor (#683) fix(deepagents): make memory permissive if AGENTS.md not found (#682)

View release on GitHub
deepagents==0.3.2 Security relevant
Security fixes
  • CVE-2025-68664
Notable features
  • memory system for agent state management
  • skills SDK support
  • improved async support in skills
View release on GitHub

Beta — feedback welcome: [email protected]