The AI Toolchain — issue -187, February 11, 2026

Glamourous agentic coding for all

Crush v0.42.0 adds GLM 5 via z.AI and surfaces user-configured LSP servers in the UI before they start.

$ git clone --branch v0.42.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:
$ git checkout v0.42.0

• ›Adds GLM 5 model support via z.AI as a new provider option.
• ›Shows user-configured LSP servers in the UI with an 'unstarted' status before they are invoked.

GitHub Copilot CLI v0.0.407 adds /on-air mode, --alt-screen, --resume with UUID, /instructions, workspace-local MCP via .vscode/mcp.json, and a tools.list RPC.

$ git clone --branch v0.0.407 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:
$ git checkout v0.0.407

{
  "servers": {
    "my-mcp-server": {
      "command": "npx",
      "args": ["my-mcp-package"]
    }
  }
}

• ›Adds --alt-screen flag to enable experimental alternate screen buffer mode.
• ›Adds --resume flag that creates new sessions with a provided UUID.
• ›Adds workspace-local MCP server configuration via .vscode/mcp.json.
• ›Adds tools.list RPC to query all available built-in tools at runtime.
• ›Adds /on-air mode to hide model names and quota details during streaming sessions.

Lightweight coding agent that runs in your terminal

Codex CLI v0.99.0 adds concurrent shell execution, new TUI controls, app-server steering APIs, and GIF/WebP image support.

$ git clone --branch rust-v0.99.0 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.99.0

$ /statusline

# requirements.toml
[requirements]
allowed_web_search_modes = ["safe"]

[network]
# network constraints defined here

• ›Enables concurrent shell command execution alongside active agent turns without interruption.
• ›Adds /statusline command for interactively configuring which metadata appears in the TUI footer.
• ›Adds sortable resume picker with toggle between creation time and last-updated time, with an in-picker mode indicator.
• ›New app-server APIs for steering active turns (turn/steer), listing experimental features, resuming agents (resume_agent), and opting out of specific notifications.
• ›Enterprise requirements.toml now supports restricting allowed web search modes (allowed_web_search_modes) and defining network constraints.

The open source coding agent.

OpenCode v1.1.58 adds mode-specific input placeholders, Ctrl+C dialog dismissal, and persistent /share link access in the TUI.

$ git clone --branch v1.1.58 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.1.58

• ›Adds mode-specific input placeholders in the TUI prompt for context-aware guidance depending on the active mode.
• ›Enables dismissing dialogs with Ctrl+C in the TUI.
• ›Keeps the /share command available after sharing so users can copy an existing share link at any time.
• ›Enables notifications on child sessions in the desktop app.

$ git clone --branch v1.1.57 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.1.57

Qwen Code v0.10.1 adds MCP tool progress updates and a new Coding Plan authentication mode with configuration management.

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

• ›Adds Coding Plan configuration management, introducing a new authentication mode with a unified AuthDialog for Coding Plan workflows.
• ›Adds MCP tool progress update support in TUI and SDK mode, surfacing real-time progress for long-running MCP tool calls.

$ git clone --branch v0.10.0-preview.6 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.10.0-preview.6

Zed v0.223.3 adds Agent URL-based prompting, terminal-to-Agent selections, dangerous command blocking, Azure Git, and remote file downloads.

$ git clone --branch v0.223.3 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.223.3

$ open 'zed://agent?prompt=Review%20the%20latest%20diff%20for%20secrets%20or%20hardcoded%20credentials'

"working_directory": "current_file_directory"

• ›Opens the Agent Panel with a pre-filled prompt via zed://agent?prompt=<url_encoded_text> URL, enabling deep-links into AI-assisted workflows.
• ›Adds current_file_directory as a valid value for the terminal working_directory setting ("working_directory": "current_file_directory") to open new terminals in the active file's directory.
• ›Adds a settings page in the settings editor for customizing tool-call permissions per Agent tool.
• ›Adds support for ACP Session Config options for agents that provide them.
• ›Adds automatic blocking of known-dangerous Agent terminal commands such as rm -rf / and rm -rf ~.

Jan v0.7.7 adds MLX backend for Apple Silicon, Exa MCP search, vision model prompts, and local API server improvements.

$ git clone --branch v0.7.7 https://github.com/janhq/jan.git
# already have the repo? check out this version:
$ git checkout v0.7.7

• ›Adds new MLX backend for running inference natively on Apple Silicon Macs.
• ›Adds streamable HTTP Exa default search MCP integration.
• ›Adds prompt to download the Jan VL model when vision capability is needed.
• ›Adds 'delete all threads' action inside projects.
• ›Implements auto-fit settings in the llama.cpp extension to automatically size models to available resources, replacing the previous model planner.

OpenAI Agents SDK v0.8.4 adds ShellTool with container runtime and native skills support.

$ git clone --branch v0.8.4 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.8.4

from agents import Agent, ShellTool

agent = Agent(
    name="Shell Agent",
    model="gpt-5.2",
    instructions="Use the available shell tool to answer user requests.",
    tools=[
        ShellTool(
            environment={
                "type": "container_auto",
                "network_policy": {"type": "disabled"},
                "skills": [
                    {
                        "type": "skill_reference",
                        "skill_id": "skill_698bbe879adc81918725cbc69dcae7960bc5613dadaed377",
                        "version": "1",
                    }
                ],
            }
        )
    ],
)

• ›Adds ShellTool with environment parameter supporting type: container_auto, network_policy, and skills (via skill_reference with skill_id and version) for hosted container shell runtime with native skills support.

PydanticAI v1.58.0 adds report-level evaluators, multi-run aggregation, and extra_headers for Google provider

$ git clone --branch v1.58.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v1.58.0

from pydantic_ai.models.google import GoogleModel

model = GoogleModel('gemini-2.0-flash', extra_headers={'X-My-Trace-Id': 'abc123'})

from pydantic_evals import Dataset

results = await dataset.evaluate(pipeline, repeat=5)

• ›Adds extra_headers support to the Google model provider, enabling custom HTTP headers on requests.
• ›Adds a repeat parameter to pydantic-evals for multi-run aggregation, enabling statistical analysis across repeated experiment runs.
• ›Introduces report-level evaluators and experiment-wide analyses to pydantic-evals, enabling summary metrics across all cases in an evaluation run.