other / agent
Pizza Bot
An inbox for long-running AI agent work, with desktop, browser and terminal clients, approval requests and scheduled tasks.
At a glance
- Purpose
- An inbox for long-running AI agent work, with desktop, browser and terminal clients, approval requests and scheduled tasks.
- Runs in
- Local agent with shell, Web service
- Operating systems
- macOS, Linux, Windows
- Needs
- Runtime: Node.js 24+; Needs internet: Not stated; Needs accounts or API keys beyond your agent: Not stated
- Files and data
- Reads files: Yes; Writes files: Yes; Sends data to other services: Yes
- Tool type
- Agent
- Delivery
- From GitHub
- Version
- Latest on the main branch
- Access
- Code on GitHub
- Compatibility
- According to the README: macOS (Apple silicon and Intel) with .dmg or .zip packages, Windows (x64) with a -setup.exe installer, and Linux (x64 and arm64) with .deb or .rpm packages. The README requires Node.js 24 or newer for source use and lists Electron desktop, browser and terminal CLI clients. The README points to a standalone backend guide for authentication, remote Electron, static browser deployment, Docker, Compose and Kubernetes. The README lists Amazon Bedrock, Anthropic, Google Gemini, OpenAI, OpenRouter, Requesty and Ollama as model providers, and describes MCP servers, Agent Skills and plugins.
- External costs
- According to the README, you bring a model provider: Amazon Bedrock, Anthropic, Google Gemini, OpenAI, OpenRouter, Requesty or Ollama. Bedrock accepts an AWS profile, AWS access keys or a Bedrock API key. The README lists no prices. The provider you connect may charge for usage.
- Permissions
- The README's source route installs dependencies with npm install, builds the app and starts an Electron shell that supervises an api-server. The README says application state defaults to ~/.pizza-bot-oss, with MCP configuration and skills under the data root. The README requires configured model access and describes AWS profile, access-key or Bedrock API-key authentication. The README says local folders require explicit read-only or writable grants, and remote grants refer to backend-host paths. The README describes cron and webhook triggers, long-running tasks and human-in-the-loop approval requests. The README says the api-server defaults to 127.0.0.1 and non-loopback binding requires authentication and an origin allowlist. The README warns that MCP-server and plugin commands and materializers can execute with the user's account permissions. The top-level file list includes scripts/, whose contents the README does not describe.
- Usage terms
- Apache License 2.0, as stated in the repository. See its LICENSE file for the exact terms. The README also points to a NOTICE file for attribution notices.
- Interface
- desktop
Code on GitHub
pizza-bot-app/pizza-bot
Apache-2.0A local-first inbox for long-running AI agents, built with DeepAgents and LangGraph.
Topics: agentic-ai ai-agents amazon-bedrock deepagents electron human-in-the-loop langgraph local-first
Public GitHub data fetched · 0 days old
Skill folders in this repository (3)
Folders that contain a SKILL.md file, from GitHub's file list on .
What it does
According to the README, Pizza Bot is an inbox for long-running AI work, with finished work in Unread and approval requests in Action. The README says agents continue after client navigation or disconnection while the api-server remains running.The README describes a stateful DeepAgents and LangGraph runtime serving Electron and browser clients, with desktop, browser and terminal clients connecting over HTTP/SSE. The README says checkpointed runs survive client disconnects and cron or webhook triggers can start work. The README describes thread folders and skills becoming tool-scoped subagents with Activity-panel progress.
The README describes MCP servers, Agent Skills, plugins and explicit read-only or writable local-folder grants. The README lists Amazon Bedrock, Anthropic, Google Gemini, OpenAI, OpenRouter, Requesty and Ollama as model providers.
Who it is for
People who want to start or schedule long-running AI agent tasks and read the results and approval requests later in a desktop app, a browser or a terminal.Example tasks
Start a task and read the finished work in the Unread queue.Answer an approval request waiting in the Action queue.
Schedule a task or start one from a cron or webhook trigger.
Grant a local folder under Settings > Files as read-only or writable.
Add an MCP server or an Agent Skill.