Skip to content

One-Paste Installs

These are the cleanest client wiring patterns for SkillForge. The public repo is now skillforge, while the installable runtime package remains orchestrator-mcp. Use the same Python executable that installed the package. If your machine exposes python3 instead of python, swap that value in every snippet below.

Codex

bash
codex mcp add orchestrator -- python -m orchestrator_mcp.cli serve

Config file alternative:

toml
[mcp_servers.orchestrator]
command = "python"
args = ["-m", "orchestrator_mcp.cli", "serve"]

Claude Code

bash
claude mcp add orchestrator -- python -m orchestrator_mcp.cli serve

Project config example:

json
{
  "mcpServers": {
    "orchestrator": {
      "command": "python",
      "args": ["-m", "orchestrator_mcp.cli", "serve"]
    }
  }
}

Kimi Code

Kimi Code documents local MCP setup through its MCP Servers UI. Use:

  • Transport: stdio
  • Command: python
  • Args: -m orchestrator_mcp.cli serve

The repo also includes a generic stdio example in examples/kimi-stdio.json.

Generic MCP client

Any client that can launch an stdio MCP server can use the same pattern:

bash
python -m orchestrator_mcp.cli serve

After wiring it in

Once the client can start the runtime, validate the broader setup with:

bash
orchestrator-mcp doctor

Then browse Marketplace if you want the portable packs, or head back to Quickstart if you still need the runtime path.