PocketPaw AI Agent

Description

PocketPaw redefines the AI agent experience by prioritizing modularity, security, and universal access. Unlike traditional agents that require extensive setup, Docker, and complex configurations, PocketPaw offers a streamlined 30-second installation process. This open-source solution, built with Python and licensed under MIT, allows users to interact with their AI agent through familiar platforms like Telegram, Discord, Slack, and WhatsApp, as well as a dedicated web dashboard.

The core problem PocketPaw addresses is the time-consuming and often insecure setup of AI agents. Typical setups involve provisioning VMs, SSH authentication, cloning repositories, editing numerous configuration variables, setting up OAuth, tunneling, and installing hundreds of dependencies, often leading to conflicts and instability. PocketPaw bypasses these hurdles with a single command installation that immediately provides a running dashboard, encrypted credentials, connected channels, and initialized memory, ready for the first message.

PocketPaw's modular design means users only install what they need. The 'batteries included' approach offers a comprehensive package with a web dashboard, command center, Telegram bot, multiple LLM backends (Claude Agent SDK, OpenAI, Ollama), encrypted credentials, persistent memory, browser automation, and an audit log. Users can opt for a thin core installation and add specific functionalities like semantic search via Mem0 or install all features with `pip install pocketpaw [all]`.

Security is a foundational aspect of PocketPaw. Credentials are encrypted using Fernet AES with a machine-derived PBKDF2 key, moving beyond simple base64 obfuscation. A Guardian AI feature acts as a secondary LLM to review dangerous commands before execution, supported by a 3-tier tool policy with deny-lists. The agent operates with no exposed ports by default, prioritizing Telegram for initial interaction, and uses WebSocket authentication via message body to prevent exposure in logs or scan results. An append-only, tamper-evident audit log records every tool execution.

The Command Center feature enhances productivity by enabling teams to manage complex goals. It breaks down objectives into smaller tasks, assigns agents, streams progress live, and requires user approval before executing critical actions. This facilitates collaborative workflows for tasks like content pipelines, bug fixing, guest bookings, and security audits.

PocketPaw is designed for users who want control over their AI agents, data privacy, and operational costs. Its open-source nature, MIT license, and lack of telemetry or cloud dependency ensure users retain full ownership and autonomy. The ability to run local LLM models via Ollama further reduces API costs, making it an economical choice for advanced AI interactions.

PocketPaw AI Agent's Core Features

• Self-hosted AI agent

• Open-source with MIT license

• Modular installation via pip extras

• Multi-channel support (Telegram, Discord, Slack, WhatsApp, Web Dashboard)

• Multiple LLM backend options (Claude Agent SDK, OpenAI, Ollama)

• Encrypted credentials with Fernet AES

• Guardian AI for command safety checks

• Persistent memory with optional semantic search (Mem0)

• Browser automation capabilities

• Append-only, tamper-evident audit log

• Command Center for task management and collaboration

• Fast 30-second installation

• No Docker or tunneling required

• Low memory footprint (~80 MB for thin core)

How to use PocketPaw AI Agent?

1. 1

   Install: Run `curl -fsSL https://pocketpaw.xyz/install.sh | sh` or `pip install pocketpaw`.

2. 2

   Configure LLM: Set up your preferred LLM backend (Ollama, OpenAI, Claude Agent SDK).

3. 3

   Connect Channels: Add desired messaging platforms via the dashboard or configuration.

4. 4

   Initialize Memory: Start with session history or upgrade to semantic search.

5. 5

   Interact: Send your first message through any connected channel.

6. 6

   Automate Tasks: Utilize built-in tools and the Command Center for complex workflows.

7. 7

   Extend Capabilities: Add new skills and tools from a registry or build your own.

PocketPaw AI Agent's Use Cases

• Personal Assistant
• Code Generation
• Web Automation
• Content Creation
• Team Collaboration
• Data Analysis
• Security Auditing

FAQ from PocketPaw AI Agent

• What is a self-hosted AI agent?

  A self-hosted AI agent runs on your own hardware, not in the cloud. It can perform actions like answering questions, executing code, browsing the web, and managing files, with your data remaining on your machine. PocketPaw is an open-source example connecting to various messaging platforms.

• Can I run PocketPaw without an API key?

  Yes, PocketPaw supports Ollama, which allows you to run open-source LLMs locally on your hardware. This eliminates the need for API keys and associated usage fees. A minimum of 4GB RAM is recommended for a 7B parameter model.

• Which messaging platforms does PocketPaw support?

  PocketPaw connects to over 9 platforms, including Telegram, Discord, Slack, WhatsApp, Signal, Matrix, Microsoft Teams, Google Chat, and a built-in web dashboard. All these channels share the same AI agent, tools, and conversation memory.

• Is PocketPaw free to use?

  PocketPaw is completely free and open-source under the MIT license. If you choose to use cloud-based LLM providers like Anthropic or OpenAI, their standard API pricing will apply. Using Ollama with local models offers a zero-cost operation.

• What is the difference between an AI agent and a chatbot?

  A chatbot primarily generates text responses. An AI agent, like PocketPaw, can take real-world actions such as browsing websites, writing code, managing files, sending emails, and completing multi-step tasks. PocketPaw includes over 50 built-in tools for these actions.

• How do I install PocketPaw?

  Installation is straightforward. You can use the interactive installer script by running `curl -fsSL https://pocketpaw.xyz/install.sh | sh`, or install it directly using pip with `pip install pocketpaw`. The process takes under 5 minutes and requires Python 3.11+.

• How does PocketPaw ensure security?

  PocketPaw employs robust security measures including Fernet AES encrypted credentials, a machine-derived PBKDF2 key, a 3-tier tool policy with deny-lists, and a Guardian AI safety check for dangerous commands. It also features an append-only, tamper-evident audit log and no exposed ports by default.

• What are the system requirements for running PocketPaw locally?

  For local LLM inference with Ollama, a minimum of 4GB of RAM is recommended for a 7B parameter model. The core PocketPaw installation has a small footprint of approximately 80 MB.

• Can PocketPaw manage complex, multi-step tasks?

  Yes, PocketPaw's Command Center is designed to break down complex goals into smaller tasks, assign them to agents, stream progress, and pause for user approval. This facilitates collaborative workflows for various projects.

• What LLM providers are supported by PocketPaw?

  PocketPaw supports Claude Agent SDK, OpenAI, and Ollama for fully local models. This flexibility allows users to choose between cloud-based services or cost-effective local inference.