The AI Toolchain — issue -244, December 16, 2025

Autonomous coding agent as an SDK, IDE extension, or CLI assistant.

Cline v3.43.1 adds GLM-4.6 and kat-coder-pro models plus env variable expansion in mcpconfig.json

$ git clone --branch v3.43.1 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.43.1

• ›Adds GLM-4.6 model support via OpenRouter.
• ›Adds kat-coder-pro model support.
• ›Supports parsing of environment variable patterns inside mcpconfig.json, enabling dynamic configuration via env vars.

Glamourous agentic coding for all

Crush v0.26.0 adds Hyper terminal integration.

$ git clone --branch v0.26.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:
$ git checkout v0.26.0

• ›Adds Hyper terminal integration.

$ git clone --branch v0.25.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:
$ git checkout v0.25.0

$ crush projects --json

The open source coding agent.

OpenCode v1.0.164 adds experimental oxfmt formatter, shell alias expansion, and desktop session/environment improvements.

$ git clone --branch v1.0.164 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.0.164

$ ! gs

• ›Adds experimental oxfmt formatter support for code formatting.
• ›Enables shell alias expansion when using the ! command in the TUI.
• ›Adds a button in the desktop app to load more sessions per project.
• ›Desktop app now loads user shell environment variables.
• ›Desktop app now shows retry attempts during tool execution.

OpenHands: AI-Driven Development

OpenHands 1.0.0 adds Azure DevOps integration, a new task tracker UI, and native tool calling for Gemini 2.5 Pro.

$ git clone --branch 1.0.0 https://github.com/All-Hands-AI/OpenHands.git
# already have the repo? check out this version:
$ git checkout 1.0.0

• ›Adds Azure DevOps integration support for repository and pipeline connectivity.
• ›New task tracker interface provides structured visibility into agent task progress.
• ›Chat messages are now queued during runtime connection, enabling smoother interaction during agent startup.

AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI

Pi v0.22.4 adds --list-models to browse available models with fuzzy search and capability details.

$ git clone --branch v0.22.4 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:
$ git checkout v0.22.4

$ pi --list-models image

• ›New --list-models [search] flag lists all available models filtered by configured API keys, showing provider, model ID, context window, max output, thinking support, and image support — with optional fuzzy search.

Qwen Code v0.5.1 adds a chat recording toggle accessible via CLI and settings.

$ git clone --branch v0.5.1-nightly.20251216.52faa0da https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.5.1-nightly.20251216.52faa0da

• ›Adds a chat recording toggle available via both CLI flag and the settings interface, with recording automatically disabled during tests.

Zed v0.217.1 adds command palette history, git remotes, Grok 4.1 Fast with 2M-token context, and new multi-cursor rotation actions.

$ git clone --branch v0.217.1 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.217.1

{
  "preview_tabs": {
    "enable_preview_from_project_panel": true,
    "enable_preview_from_file_finder": false,
    "enable_preview_from_multibuffer": true,
    "enable_preview_multibuffer_from_code_navigation": false,
    "enable_preview_file_from_code_navigation": true,
    "enable_keep_preview_on_code_navigation": false
  }
}

{
  "language_models": {
    "bedrock": {
      "allow_global": true
    }
  }
}

• ›Adds RotateSelectionsForward and RotateSelectionsBackward actions to rotate content in a circular fashion across multiple cursors.
• ›Adds editor::InsertSnippet action for programmatic snippet insertion.
• ›Adds preview_tabs.enable_preview_from_project_panel (default true), preview_tabs.enable_preview_from_multibuffer (default true), preview_tabs.enable_preview_multibuffer_from_code_navigation (default false), and preview_tabs.enable_preview_file_from_code_navigation (default true) settings as part of an overhauled preview tabs configuration.
• ›Adds opt-in allow_global config key for the Bedrock provider to enable global endpoints.
• ›Adds command palette history — pressing up now cycles through recently executed commands, useful for re-running near-miss vim-mode commands.

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

gemini-cli v0.21.0 adds hook lifecycle events, MCP Resources, OSC52 remote copy, extension settings commands, and more.

$ git clone --branch v0.21.0 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.21.0

$ gemini
# inside the session:
/copy

$ gemini mcp add my-server --type sse --url https://my-mcp-server.example.com/sse

• ›Adds hook system for session lifecycle and compression integration, enabling scripts to react to session start/end and context compression events.
• ›Adds a Hooks Commands Panel in the UI with per-hook enable/disable toggles and migration support.
• ›Supports Extension Hooks with a security warning when a hook is loaded from an extension.
• ›Adds --enableAgents experimental flag to opt into agent-based workflows.
• ›Supports MCP Resources, allowing MCP servers to expose file/data resources to the CLI.

Framework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.

CrewAI 1.7.1 adds a --no-commit flag to the bump command and switches to JSON schema for tool argument serialization.

$ git clone --branch 1.7.1 https://github.com/crewAIInc/crewAI.git
# already have the repo? check out this version:
$ git checkout 1.7.1

$ crewai bump --no-commit

• ›Adds --no-commit flag to the bump command, allowing version bumps without auto-committing changes.
• ›Switches tool argument serialization to use JSON schema for more structured, interoperable tool definitions.

PydanticAI v1.34.0 adds a Web Chat UI launchable via clai web or Agent.to_web(), plus FileUrl.force_download support for Anthropic and OpenAI Responses models.

$ git clone --branch v1.34.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v1.34.0

$ clai web

from pydantic_ai import Agent

agent = Agent('openai:gpt-4o', system_prompt='You are a helpful assistant.')
agent.to_web()

• ›Adds clai web CLI command and Agent.to_web() method to launch a Web Chat UI for any agent.
• ›Supports FileUrl.force_download in AnthropicModel and OpenAIResponsesModel for forced file downloads.
• ›Makes OpenRouterProvider and DeepSeekProvider __init__ overloads less restrictive, broadening valid initialization patterns.

$ git clone --branch v1.33.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v1.33.0

browser-use 0.11.2 adds support for the new BU OSS model and introduces a max-wait cap between action executions.

$ git clone --branch 0.11.2 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.11.2

• ›Adds support for the open-source browser-use/bu-30b-a3b-preview model (hosted on Hugging Face) as a first-party option for driving browser automation.
• ›Introduces a maximum wait interval between action executions, capping idle time between browser steps.

llama.cpp b7429 adds GLM4V vision encoder support including Metal acceleration and MoE variant.

$ git clone --branch b7429 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b7429

• ›Adds support for the GLM4V vision encoder model, including Metal (Apple GPU) acceleration and MoE variant support.
• ›Introduces GGML_ROPE_TYPE_MRNORM rope type to support the GLM4V text model's multi-resolution positional encoding.

$ git clone --branch b7426 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b7426

$ git clone --branch b7418 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b7418

$ git clone --branch b7414 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b7414

Composio v0.10.1 adds composio generate --toolkits flag for scoped code generation in TypeScript and Python.

$ git clone --branch v0.10.1 https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:
$ git checkout v0.10.1

• ›Adds --toolkits [toolkits] argument to composio generate, letting practitioners scope code generation to specific toolkits for both TypeScript and Python.

LanceDB v0.23.0 adds IVF SQ index, async namespace connections, to_pydantic async support, stable row IDs via storage_options, and head() for remote tables.

$ git clone --branch v0.23.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout v0.23.0

table = db.create_table('my_table', data=df, storage_options={'stable_row_ids': 'true'})

• ›Adds storage_options support for enabling stable row IDs on tables.
• ›Adds num_attempts field to merge insert results, giving callers visibility into retry behaviour.
• ›Adds to_pydantic support in the async Python API for converting query results directly to Pydantic models.
• ›Implements head() for remote tables, enabling fast row-count-limited fetches against remote backends.
• ›Adds IVF SQ (Scalar Quantization) index support and HNSW aliases for index creation.

$ git clone --branch python-v0.26.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout python-v0.26.0

import lancedb

db = lancedb.connect("<uri>")
tbl = db.create_table(
    "my_table",
    data=my_data,
    storage_options={"stable_row_ids": "true"}
)

import lancedb
from pydantic import BaseModel

class Item(BaseModel):
    id: int
    text: str
    vector: list[float]

async def query():
    db = await lancedb.connect_async("<uri>")
    tbl = await db.open_table("my_table")
    results = await tbl.query().limit(10).to_pydantic(Item)
    return results

import lancedb

db = lancedb.connect("<uri>")
tbl = db.open_table("my_table")
tbl.create_index(metric="cosine", index_type="IVF_SQ", vector_column_name="vector")