The AI Toolchain — issue -341, September 9, 2025

Glamourous agentic coding for all

Crush v0.7.8 adds multiline paste support on Windows and enables SQLite secure_delete for safer data handling.

$ git clone --branch v0.7.8 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:
$ git checkout v0.7.8

• ›Enables bracketed paste on Windows, allowing multiline text to be pasted correctly.
• ›Enables SQLite secure_delete pragma so deleted database records are securely overwritten.

OpenHands: AI-Driven Development

OpenHands 0.56.0 adds AGENTS.md support, microagent management, and multi-LLM routing within agent trajectories.

$ git clone --branch 0.56.0 https://github.com/All-Hands-AI/OpenHands.git
# already have the repo? check out this version:
$ git checkout 0.56.0

• ›Supports AGENTS.md files in microagents, enabling repo-level agent behavior configuration via a standardized file.
• ›New microagent management feature for organizing and controlling microagents.
• ›Enables routing between different LLMs within a single agent trajectory, allowing dynamic model selection per task step.

Qwen Code v0.0.11-nightly.2 adds configurable cache control and corrected token limits for Qwen models.

$ git clone --branch v0.0.11-nightly.2 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.0.11-nightly.2

• ›Adds configurable cache control via new configuration support, enabling tuning of caching behavior.
• ›Re-implements the tokenLimits class to work correctly for Qwen and related models, unlocking accurate context-window management.

Zed v0.202.8 passes the project shell environment to external agent processes.

$ git clone --branch v0.202.8 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.202.8

• ›Agent processes now receive the project shell environment, enabling external agents to inherit the same env vars as the editor session.

An open-source AI agent that brings the power of Gemini directly into your terminal.

gemini-cli v0.4.0 adds session recording, smart edits, fuzzy command matching, citations, extension linking, and more new CLI capabilities.

$ git clone --branch v0.4.0 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.4.0

$ gemini --session-summary

$ gemini extensions link <path>

$ gemini "Summarize the changes in the last 5 git commits"

• ›Adds --session-summary flag to generate a summary at the end of a session.
• ›Adds gemini extensions link command to link local extensions for development.
• ›Adds flag to update all extensions at once.
• ›Supports installing extensions using org/repo shorthand syntax.
• ›Adds fuzzy matching for command suggestions when typing / commands.

$ git clone --branch v0.4.0-preview.4 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.4.0-preview.4

$ gemini --session-summary

$ gemini extensions link /path/to/my-extension

$ gemini "Summarize the changes in CHANGELOG.md"

$ git clone --branch v0.5.0-nightly.20250909.2b05cf3b https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.5.0-nightly.20250909.2b05cf3b

langchain-groq 0.3.8 adds json_schema structured-output support for Groq models.

$ git clone --branch langchain-groq==0.3.8 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-groq==0.3.8

• ›Adds support for json_schema structured output mode when calling Groq models, enabling strict schema-constrained responses.

PydanticAI v1.0.2 adds DBOS durable execution, sequential tool calling, and Google cached content support

$ git clone --branch v1.0.2 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v1.0.2

from pydantic_ai.models.google import GoogleModelSettings

settings = GoogleModelSettings(
    google_cached_content="cachedContents/abc123"
)
result = await agent.run("Summarize the document.", model_settings=settings)

• ›Adds GoogleModelSettings.google_cached_content field to pass cached_content when calling Google models, enabling prompt caching.
• ›Adds ModelResponse.finish_reason attribute and populates provider_response_id during streaming responses.
• ›Adds support for gen_ai.response.id in OpenTelemetry instrumentation spans.
• ›Adds support for durable execution with DBOS, enabling fault-tolerant, resumable agent workflows.
• ›Adds support for sequential tool calling, allowing agents to invoke tools one at a time in order rather than in parallel.

browser-use 0.7.7 adds an enhanced screenshot API and automatic PDF download detection via DownloadsWatchdog.

$ git clone --branch 0.7.7 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.7.7

• ›Adds an enhanced screenshot API with new capabilities for capturing browser state programmatically.
• ›Extends DownloadsWatchdog to watch BrowserStateRequestEvent and automatically handle PDF downloads.
• ›Removes the hardcoded maximum iframe recursion limit, enabling DOM traversal into arbitrarily nested iframes.