Progi

io.github.zsetaPublished by GitHub account zsetav0.3.2Updated Oct 24, 2026

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

VerifiedSTDIODesktop onlyProductivity & WorkflowKnowledge & Memory

Overview

AI-generated overview

Progi is a local MCP workflow engine that stores your processes as structured workflows and guides an assistant through their steps.

What you can do
Progi lets an assistant turn a plain-language description of a process into a stored workflow with per-step playbooks, then run tasks against it. Tools include create_task, list_tasks, start_or_continue_task, update_progress_notes and finish_step for the work loop, plus get_process_skeleton_prompt, get_playbook_authoring_prompt, save_workflow and list_workflows for authoring. Workflows and task state live in a SQLite database, so they survive context resets. A local web UI shows running and completed tasks with status, progress and output history.
When to use it
Useful when you repeat multi-step processes and want an assistant to follow your own playbooks instead of re-explaining them each session. Suited to workflows with checkpoints where you review output mid-run. Less useful for one-off tasks with no repeatable structure.
Requirements
Runs locally over stdio, typically via uvx progi, so Python and uv are needed. No account or API key is declared. Optional environment variables: PROGI_DB_PATH for the SQLite file (use an absolute path), PROGI_WEB_HOST, PROGI_WEB_PORT and PROGI_NO_WEB for the local web UI. Desktop only; no web executable.
Before you install
The server writes workflow, task and output data to a local SQLite database, so point PROGI_DB_PATH at a location you are comfortable persisting. A local web UI is started by default on 127.0.0.1:8000; it can be disabled with PROGI_NO_WEB or bound elsewhere with PROGI_WEB_HOST. Task outputs and progress notes are stored, so avoid putting secrets in them.

Installation

In SourceWeft

  1. Open Progi in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Progi - MCP-native Workflow Engine

[Progi]

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.

[License: MIT] [PyPI] [MCP]

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

ToolDescription
create_taskCreate a new task under a given workflow (status todo); returns a preview of its first step
list_tasksList tasks, optionally filtered by status and/or workflow
start_or_continue_taskMain work-loop entry point — starts or resumes a task and returns the current step's playbook, input data, and output spec
update_progress_notesOverwrite a task's progress notes (mid-step save point)
finish_stepMark the current step complete, store its output, and advance to the next step (or mark done)

Workflow authoring

ToolDescription
get_process_skeleton_promptReturn the Pass 1 system prompt for turning a plain-language description into a structured workflow skeleton
get_playbook_authoring_promptReturn the Pass 2 system prompt for authoring a step's playbook (injects workflow context)
save_workflowPersist a new workflow, its steps, and playbooks
list_workflowsReturn 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

VariableDefaultPurpose
PROGI_DB_PATHOS data dir (platformdirs)SQLite file location
PROGI_WEB_HOST127.0.0.1Web UI bind host
PROGI_WEB_PORT8000Web UI port
PROGI_NO_WEB0Set 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

Source: README.md at commit 82753f9

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.3.2LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts