PET-P

中文 | English

A techno-person RPA toolkit — configurable task runner, execution engine, and MCP Tool Server built with Python.
Friendly for AI Agents, DevOps, and automation testing.

Project Sites: PETP Intro | Web App Guide

PET = Pipeline-Execution-Task, the hierarchical execution model. The trailing P stands for Processor, which handles each task one-to-one.

Pipeline  1:n Execution
Execution 1:n Task
Task      1:1 Processor

Table of Contents

• Features
• Screenshots
• Quick Start
• Dependency Management
• Running Modes
• HTTP Service & MCP
• Build & Docker
• Web App (Docker & UGOS)
• Project Structure
• Troubleshooting
• Acknowledgements
• Changelog

Features

Orchestrate tasks as Executions (with dataset-loop or time-loop), combine Executions as Pipelines, run once or on cron.

Screenshots

macOS

Windows

HTTP Service Enabled

Run first Execution within 4 steps:

LLM MCP Server, Client & Host:

MCP Tool Server (Streamable-HTTP):

Integrate with Claude Code:

Quick Start

Prerequisites

ChromeDriver location: webdriver/darwin/chromedriver (macOS) or webdriver/win32/chromedriver.exe (Windows)

Step 1 — Install Python

Download and install Python 3.14 from python.org.

On Windows, check "Add Python to PATH" during installation.

Step 2 — Install wxPython

wxPython must match your exact Python version and OS. Download the .whl from wxpython.org/Phoenix/snapshot-builds:

Always download the latest snapshot for best compatibility.

Step 3 — Install Dependencies

See Dependency Management for full details. Quick install:

# Recommended: install with uv
pip install -U uv

# If you see: "No virtual environment found"
uv venv
# Creating virtual environment at: .venv
# Activate with: .venv\Scripts\activate

uv pip install -r requirements.txt

# Or install only what you need (see [Dependency Groups](#dependency-groups))
uv pip install -r requirements/core.txt -r requirements/ssh-sftp.txt

# Alternative: install with pip
pip install -r requirements.txt

# Or install only what you need (see [Dependency Groups](#dependency-groups))
pip install -r requirements/core.txt -r requirements/ssh-sftp.txt

Step 4 — Run

python PETP.py

The GUI launches. On first run, a default config is created at ./config/petpconfig.yaml.

macOS helper scripts (recommended for long-running sessions)

chmod +x scripts/macos/start_petp.sh scripts/macos/start_petp_gui.sh scripts/macos/start_petp_background.sh

# Unified launcher (recommended)
./scripts/macos/start_petp.sh gui
./scripts/macos/start_petp.sh bg --run-execution ENDECODER --no-http

# Legacy wrappers (still supported)
./scripts/macos/start_petp_gui.sh
./scripts/macos/start_petp_background.sh --run-execution ENDECODER --no-http

Defaults used by both scripts (only when not already set):

• PYTHONMALLOC=malloc
• PYTHONUNBUFFERED=1
• PYTHONDONTWRITEBYTECODE=1

Optional overrides:

PYTHON_BIN=python3.14 PETP_ECHO_ENV=1 ./scripts/macos/start_petp.sh gui
PYTHONMALLOC=malloc ./scripts/macos/start_petp.sh background --run-pipeline MY_PIPELINE --no-http

Windows helper scripts (recommended for long-running sessions)

First-time setup — allow PowerShell scripts to run (once per machine):

Set-ExecutionPolicy -Scope CurrentUser RemoteSigned

# Unified launcher (recommended)
.\scripts\windows\start_petp.ps1 gui
.\scripts\windows\start_petp.ps1 bg --run-execution ENDECODER --no-http

# Dedicated wrappers
.\scripts\windows\start_petp_gui.ps1
.\scripts\windows\start_petp_background.ps1 --run-execution ENDECODER --no-http

# Show help
.\scripts\windows\start_petp.ps1 help

Defaults applied automatically (only when not already set):

• PYTHONUNBUFFERED=1
• PYTHONDONTWRITEBYTECODE=1

Optional overrides (set env vars before calling):

$env:PYTHON_BIN = '.\.venv\Scripts\python.exe'
$env:PETP_ECHO_ENV = '1'
.\scripts\windows\start_petp.ps1 gui

# Background mode with custom Python
$env:PYTHON_BIN = 'python3.14'
.\scripts\windows\start_petp.ps1 bg --run-pipeline MY_PIPELINE --no-http

Dependency Management

PETP uses a modular dependency structure — split by processor category for flexible installation and minimal packaging.

Install with uv (Recommended)

uv is a fast Python package manager (10-100x faster than pip). Drop-in compatible with existing requirements files — zero migration needed.

# Option A (recommended): create and use a virtual environment
uv pip install -r requirements.txt

# Option B: install into system Python explicitly
uv pip install --system -r requirements.txt

# Full (GUI desktop)
uv pip install -r requirements.txt

# Background service (no GUI)
uv pip install -r requirements-nogui.txt

# Docker (headless, no browser automation)
uv pip install -r requirements-docker.txt

# Custom combination
uv pip install -r requirements/core.txt -r requirements/ssh-sftp.txt -r requirements/http-client.txt

Lock versions with uv pip compile

Generate pinned requirement files for reproducible builds:

# Compile a specific group
uv pip compile requirements/core.txt -o requirements/core.lock

# Compile the full set
uv pip compile requirements.txt -o requirements.lock

Dependency Groups

Update Dependencies

# With pip
pip list --outdated | grep -v '^\-e' | cut -d = -f 1 | xargs -n1 pip install -U

# With uv
uv pip install --upgrade -r requirements.txt

Running Modes

Background Mode

# Start background service
python PETP_backgroud.py

# Run one execution and exit
python PETP_backgroud.py --run-execution ENDECODER --no-http

Key config in petpconfig.yaml:

• nogui_enabled: true
• nogui_ui_processor_policy: skip — skip GUI processors and continue
• nogui_ui_processor_policy: abort — stop on GUI processor

Processors skipped in no-GUI mode:

Force headless outside Docker: set PETP_HEADLESS=true

HTTP Service & MCP

PETP includes a built-in HTTP server on port 8866.

For MCP Inspector: Transport Type = Streamable HTTP, URL = http://localhost:8866/mcp

Build & Docker

Build Standalone Executable

python PETP_build.py

Output: PETP.app (macOS) or PETP.exe (Windows) in ./dist/.

Docker

Fully supports building on Apple M1 (arm64) and running on x86 (amd64).

# Build & run locally
./docker_build.sh
docker run --rm -p 8866:8866 petp-background:amd64-local

# Push to registry
./docker_build.sh --push yourrepo/petp-background:1.0

Container endpoints:

Web App (Docker & UGOS)

The standalone Web App (webapp/) has its own Docker packaging guide, including UGOS (linux/amd64) build/export/import steps:

• webapp/README.md
• UGOS quick path: build linux/amd64 with buildx, verify architecture, export tar, then docker load on NAS.

Project Structure

Root Files

Directories

Troubleshooting

Acknowledgements

• wxPython & wxGlade
• Selenium & ChromeDriver & Chrome DevTool Recorder

Changelog

2026

2025

2024

2023

2022

2021