Skip to main content
Glama

Agent Hub

A modular workspace for building, organizing and orchestrating AI agents, with encapsulated skills, tools and shared capabilities.

Goals

  • Organize agents as independent modules.

  • Encapsulate agent-specific skills and tools.

  • Share common capabilities through the core.

  • Support simple and workflow-based agents.

  • Expose agents through an MCP server.

Related MCP server: SmartCLI MCP Server

Requirements

  • Python 3.12+

  • Git

Setup

Run the following commands from the project root using PowerShell:

python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install -e ".[dev]"

If PowerShell blocks script execution:

Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass
.\.venv\Scripts\Activate.ps1

Agents

The workspace currently includes:

  • Git Agent — performs Git operations in local repositories.

  • Agent Builder — assists with creating and improving agents, skills and tools.

  • Bookmark Agent — creates, extracts, describes and classifies bookmarks.

  • Project Ideas Agent — generates structured project idea documents.

Each agent has its own instructions, workflows and tools where applicable. LLM interpretation is provided through interchangeable implementations of the core Interpreter interface, currently supporting OpenAI and Anthropic.

MCP Server

The MCP server exposes the registered agents as tools over the standard input/output (stdio) transport.

Each agent is exposed through one MCP tool:

Agent

MCP tool

Git Agent

git_agent

Agent Builder

agent_builder

Bookmark Agent

bookmark_agent

Project Ideas Agent

project_ideas_agent

The server is started with:

python -m agent_hub.mcp

The current entry point configures OpenAI interpreters for all four agents. Set OPENAI_API_KEY before starting the server.

Example MCP client configuration:

{
  "mcpServers": {
    "agent-hub": {
      "command": "python",
      "args": ["-m", "agent_hub.mcp"],
      "env": {
        "OPENAI_API_KEY": "your-api-key"
      }
    }
  }
}

Adjust command to point to the Python executable in the project's virtual environment when needed.

Validation

Run the automated checks from the project root:

pytest -v
ruff check src tests
mypy src tests

All checks should pass before committing changes.

Development workflow

  1. Activate the virtual environment.

  2. Implement the changes.

  3. Run tests, lint and type checking.

  4. Commit only after all checks pass.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      D
      maintenance
      Exposes a verified tool registry (calculator, sandboxed file read, web fetch) over MCP stdio, enabling any MCP-capable client to reuse the same tools from the inspectable ReAct loop.
      3
      MIT