pocketpaw and GenericAgent

These are **competitors** — both provide self-hosted AI agents for local desktop automation and task execution, with overlapping core functionality (multi-agent orchestration, desktop control, local LLM support), though PocketPaw emphasizes ease of deployment and security while GenericAgent focuses on PC task automation loops.

pocketpaw
87
Verified
GenericAgent
60
Established
Maintenance 25/25
Adoption 17/25
Maturity 20/25
Community 25/25
Maintenance 25/25
Adoption 10/25
Maturity 3/25
Community 22/25
Stars: 601
Forks: 215
Downloads: 1,708
Commits (30d): 303
Language: Python
License: MIT
Stars: 663
Forks: 108
Downloads:
Commits (30d): 138
Language: Python
License:
No risk flags
No License No Package No Dependents

About pocketpaw

pocketpaw/pocketpaw

Your AI agent in 30 seconds. Not 30 hours. Self-hosted, open-source personal AI with desktop installer, multi-agent Command Center(Deep Work), and 7-layer security. Anthropic, OpenAI, or Ollama.

Based on the README, here's a technical summary: Integrates natively with Discord, Slack, WhatsApp, and Telegram via the web dashboard, with a cross-platform desktop app (Electron-based) bundling the Python backend and providing system tray access, global shortcuts, and side panel UI. Built on Python 3.11+ with pip distribution and Docker Compose support, featuring configurable LLM providers (Anthropic, OpenAI, Ollama) and optional vector memory persistence via Qdrant. The architecture separates a native client frontend from a self-contained backend service running on localhost:8888, enabling multi-window browsing, browser automation, and shell execution while maintaining data isolation on the user's machine.

About GenericAgent

lsdefine/GenericAgent

AI-powered PC agent loop for desktop automation and intelligent task execution

Operates through 7 atomic tools (code execution, file I/O, browser control via real session injection, web vision, ADB mobile integration) and a 92-line agent loop that autonomously crystallizes completed task execution paths into reusable skills. Supports Claude, Gemini, Kimi and other major LLMs with minimal dependencies (~3,300 lines core), progressively building a personalized skill tree from repeated task patterns without preloaded knowledge.

Scores updated daily from GitHub, PyPI, and npm data. How scores work