AgentCrew

AgentCrew is a framework for building and coordinating multiple specialized AI agents that work together to accomplish complex tasks. It exposes an agent collaboration model (A2A) over HTTP, allowing you to deploy, manage, and orchestrate teams of agents with dedicated tools and access controls. The system supports multiple AI providers (Claude, GPT, Gemini, Copilot, etc.), a rich tool ecosystem (web search, code analysis, file editing, browser automation, command execution, memory, and MCP integration), and both GUI and console interaction modes. You can run it in interactive mode for development and testing, or in job mode for automated, non-interactive tasks within pipelines.

To use MCP with AgentCrew, start the server (for example via Docker as shown in the installation guide). Once running, you can configure agents, permissions, and tool access in the GUI Settings or via configuration files. Use the A2A HTTP API to have agents from different instances collaborate on workflows, or run tasks locally in Job mode to perform single-turn analyses or transformations. Tools can be selectively enabled per agent, and tool usage can be approved or denied to maintain safety and control over automated actions.

Prerequisites

• Docker or a compatible container runtime
• Optional: Python 3.12+ if you prefer native setup (see alternatives below)
• Access keys for any preferred AI providers (Anthropic, OpenAI, Gemini, etc.)

Recommended (Docker-based) installation

1. Install Docker on your system following the official instructions for your OS.

2. Pull and run the AgentCrew Docker image:

   docker pull daltonnyx/agentcrew:latest docker run -it --rm
   -e ANTHROPIC_API_KEY="your_key"
   daltonnyx/agentcrew chat

3. If you need to customize the environment, you can pass additional environment variables to the container, such as API keys for other providers or configuration flags.

Native (Python/uv) installation (alternative)

1. Ensure Python 3.12+ is installed.

2. Install the uv package manager if you don’t have it:

   python -m pip install uv

3. Clone the repository and set up:

   git clone https://github.com/saigontechnology/AgentCrew.git cd AgentCrew uv sync uv tool install .

4. Run via uv (example):

   uv run agentcrew chat

Prerequisites summary

• Docker or Python 3.12+ with uv (if not using Docker)
• AI provider API keys (Anthropic, OpenAI, Gemini, etc.)
• Basic familiarity with environment variable configuration for keys

Tips and notes:

• The recommended primary deployment path for quick start is the Docker image ( daltonnyx/agentcrew ). It requires ANTHROPIC_API_KEY or equivalent provider keys passed as environment variables.
• Keep API keys secure and do not commit them to version control. Use environment variables or secret managers in production.
• In GUI settings or configuration files, you can define multiple agents, each with its own tool permissions and prompts. This enables targeted specialization and safer tool usage.
• If you encounter connectivity or provider errors, verify that the API keys are valid and that network access to the provider endpoints is allowed from your deployment environment.
• AgentCrew supports both interactive GUI/console modes and non-interactive Job mode, which is ideal for CI/CD pipelines and batch processing.
• When using MCP integration, ensure that cross-instance authentication (OAuth) and provider permissions are correctly configured to enable secure agent collaboration.