io.github.zseta/progi

Official

by zseta · Python

MCP-native workflow engine: teach agents your processes once, they execute them consistently.

Workflow engine that lets agents learn and consistently execute your defined processes.

Topics

claudecopilotmcpmcp-serverprogiworkflowworkflow-engineprogi-mcp

README.md4 KBView on GitHub

Progi - MCP-native Workflow Engine

Progi teaches your agent how you like to get things done. So you can do your best work without re-explaining your process or losing context between sessions.

Get started

Add Progi to your MCP client config (GH Copilot / Cursor / Claude Code / etc):

json

{
  "mcpServers": {
    "progi": {
      "command": "uvx",
      "args": ["progi"]
    }
  }
}

Progi Monitoring starts automatically at http://127.0.0.1:8000.

If you want to start Monitoring on a different port:

json

{
  "mcpServers": {
    "progi": {
      "command": "uvx",
      "args": ["progi"],
      "env": {
        "PROGI_WEB_PORT": "8080"
      }
    }
  }
}

How it works

1. Describe your workflow

"Hey Progi, help me create workflow for creating integrations, reviewing code, and publishing PRs."

Describe your process in plain language. You can be detailed or just provide a rough idea. Progi stores it as a structured workflow with per-step playbooks.

2. Run tasks, stay in the loop

"Hey Progi, start a new task, we need to review a new docs PR in the repo." Your agent loads the workflow, works through each step using your playbooks, and loops you in at critical checkpoints to review output.

3. Monitor progress

Progi Monitoring gives you a live view of every running and completed task — status, progress, and the full output history across all your workflows.

4. Optimize as you go

Tweak playbooks in Progi Monitoring between runs. Because workflows live in a database and survive context resets, every future task picks up your changes automatically — your process gets sharper with each iteration.

MCP Tools

Work loop

Tool	Description
`create_task`	Create a new task under a given workflow (status `todo`); returns a preview of its first step
`list_tasks`	List tasks, optionally filtered by status and/or workflow
`start_or_continue_task`	Main work-loop entry point — starts or resumes a task and returns the current step's playbook, input data, and output spec
`update_progress_notes`	Overwrite a task's progress notes (mid-step save point)
`finish_step`	Mark the current step complete, store its output, and advance to the next step (or mark done)

Workflow authoring

Tool	Description
`get_process_skeleton_prompt`	Return the Pass 1 system prompt for turning a plain-language description into a structured workflow skeleton
`get_playbook_authoring_prompt`	Return the Pass 2 system prompt for authoring a step's playbook (injects workflow context)
`save_workflow`	Persist a new workflow, its steps, and playbooks
`list_workflows`	Return all workflows with their ordered steps

Authoring is two passes: Pass 1 turns a plain-language description into a structured skeleton; Pass 2 authors each step's playbook. save_workflow persists both.

Configuration

Variable	Default	Purpose
`PROGI_DB_PATH`	OS data dir (`platformdirs`)	SQLite file location
`PROGI_WEB_HOST`	`127.0.0.1`	Web UI bind host
`PROGI_WEB_PORT`	`8000`	Web UI port
`PROGI_NO_WEB`	`0`	Set to `1` to disable the web UI

Run modes: uvx progi (MCP + web UI), uvx progi --no-web (MCP only), uvx progi-web (web UI only).

Use an absolute path for PROGI_DB_PATH

Install

Configuration

Environment variables

PROGI_DB_PATH

Custom path for the Progi SQLite database file

claude_desktop_config.json

{
  "mcpServers": {
    "progi": {
      "command": "uvx",
      "args": [
        "progi"
      ],
      "env": {
        "PROGI_DB_PATH": "<YOUR_PROGI_DB_PATH>"
      }
    }
  }
}