Skip to content

Agent Client Protocol

fast-agent has comprehensive support for Zed Industries Agent Client Protocol.

Why use fast-agent?:

  • Robust, native LLM Provider infrastructure, with Streaming and Structured outputs.
  • Comprehensive MCP and Agent Skills support, including Tool Progress Notifications and Sampling.
  • Build custom, multi-agent experiences in a few lines of code.

Features

Feature Support Notes
Modes ✅ Each defined Agent appears as a Modes
Tool / Workflow Progress ✅ MCP Tool Progress and Agent Workflow Progress updates
Agent Plan ✅ Iterative Planner reports progress using Agent Plan
Cancellation ✅ LLM Streaming Cancellation
Multimodal ✅ Support for Images
Slash Commands ✅ Save, Load, Status and Clear/Clear Last message
File System / Terminal ✅ Start with -x option to enable access to Client terminal
MCP Servers ⚠️ Add via command line switches or configuration file
Sessions ⚠️ Use save and load slash commands. Plan to implement with Session List

Getting Started

No Install Quick Start:

To try it out straight away with your Client, set an API Key environment variable and add:

Hugging Face

export HF_TOKEN=hf_.......

uvx fast-agent-acp@latest --model <your_model> [e.g. kimi]

Open AI

export OPENAI_API_KEY=......

uvx fast-agent-acp@latest --model <your_model> [e.g. gpt-5-mini.low]

Anthropic

export ANTHROPIC_API_KEY=......

uvx fast-agent-acp@latest --model <your_model> e.g. [sonnet]

Tip: Use uvx fast-agent-acp check to help diagnose issues.

The default system prompt will read AGENTS.md if present. Use /status system to check.

Note: OAuth keys are stored in your keyring, so check may prompt to read the credential store.

An example Zed configuration is:

...
"agent_servers": {
    "fast-agent-uvx": {
        "command": "uvx",
        "args": [
        "fast-agent-acp@latest",
        "--model",
        "kimi",
        "-x",
        "--url",
        "https://huggingface.co/mcp"
        ],
        "env": { "HF_TOKEN": "hf_xxxxxxxxxxx" }
    }
}

Installing

uv tool install -U fast-agent-mcp

The ACP Server can then be started with the fast-agent-acp command. Custom agents can be started with uv <agent.py> --transport acp.

For example:

fast-agent-acp -x --model kimi --url https://huggingface.co/mcp --auth ${HF_TOKEN}

Starts an ACP Agent, with shell access and access to the Hugging Face MCP Server.

Documentation in Progress.

Shell and File Access

fast-agent adds the read and write tools from the Client to enable "follow-along" functionality.