Player profile
OpenBot
An MIT-licensed, self-hosted platform for running a team of persistent AI bots that use tools, keep long-term memory, and hand work to each other in shared threads.
Profile
Where it fits
OpenBot is an MIT-licensed, self-hosted platform built on FastAPI, LangGraph, and LangMem, with a React web UI and an Electron desktop app for macOS and Linux. Each bot has a name, instructions, a selected set of tools, and its own long-term memory.
It belongs in Multi-Agent Platforms And Builders because the team of bots is the product surface. Bots, humans, and external systems are all actors with persistent inboxes. A bot wakes when mail arrives, runs its agent loop, and can hand the thread to one other bot with an @mention, with a hop limit to stop runaway loops. The README walks through a Chief of Staff bot that delegates to Engineer, Reviewer, and QA bots, with the QA bot asking a human for approval before it merges.
Tools include built-in shell, file, and HTTP tools rooted at a workspace directory, custom Python tools, and tools from remote or stdio MCP servers with OAuth, selectable per bot. The README notes that the shell tool is not sandboxed and that OpenBot is meant for a single trusted operator.
Builder web analytics
Measure projects built with OpenBot
OpenBot bots can write and ship changes to pages, docs, and apps through their tools. Agent Analytics can report what those changes did for real users once the changed surface is separately instrumented.
OpenBot has no built-in Agent Analytics integration. Instrument the deployed surface separately, and give the bot that should query it access through its own tools.
First loop to measure
- an OpenBot workflow, such as Engineer, Reviewer, and QA bots, ships a change to a page, docs path, or app surface
- the deployed surface, separately instrumented, reports visits, sources, signups, and configured product events to Agent Analytics
- a later bot run reads those results through an HTTP tool or MCP server
- the operator or the next bot picks the next change from the measured outcome
Copyable prompt
Use Agent Analytics for this project. If event reporting is missing, add the tracker and report events for this project surface, including OpenBot-shipped page, docs path, traffic source, signup, activation event, or funnel step. Verify events are arriving. Then fetch the last 7 days and compare them with the prior 7 days. Tell me which OpenBot-shipped page, docs path, traffic source, signup, activation event, or funnel step moved users toward value, where users dropped off, which sources mattered, and what my agent workflow should improve next.