Repolode
Submit
MarlBurroW

MarlBurroW/hivekeep

A self-hosted platform of persistent AI agents with long-term memory that collaborate and answer across chat channels, run in a single container.

View on GitHubWebsite
Stars
71
Forks
16
Open issues
26
Last commit
yesterday

What is it?

Hivekeep is a self-hosted platform for running a team of persistent AI agents on your own server. Each agent keeps one continuous session that does not reset, backed by a hybrid long-term memory that combines vector search with keyword search. Agents can message each other, delegate work to sub-agents, and run scheduled tasks.

The whole platform runs as one process with a single SQLite file in one Docker container, with no separate Postgres, Redis or queue service to provision. Users bring their own LLM API key or a local OpenAI-compatible endpoint. Secrets are stored in an AES-256-GCM vault that is not exposed to the model, and a built-in configurator agent named Queenie handles setup through conversation.

Agents can be reached through Telegram, Discord, Slack, WhatsApp, Signal, Matrix and a PWA web app, and a channel can be transferred from one agent to another mid-conversation. Agents can also create custom tools, mini-apps and plugins distributed over npm.

Key features

  • 01Persistent agents with one continuous session and hybrid long-term memory
  • 02Agents collaborate through messaging, sub-agents and scheduled tasks
  • 03Single container deployment with SQLite, no external database required
  • 04Six chat channels plus a PWA, with real-time channel handoff between agents
  • 05Secrets stored in an encrypted vault that the LLM never sees
  • 06Agents can build custom tools, mini-apps and npm-based plugins

Use cases

  • Running a personal AI assistant that remembers context across months of conversations
  • Operating a household or small team of specialist agents reachable from Telegram or Slack
  • Keeping LLM API keys and account credentials on your own server
  • Automating scheduled checks, reminders and email triage with agents

Things to know

  • The README says the project is about 80% mature and has known rough edges.
  • It is maintained by a solo developer, and much of the code is AI-written, which the README flags as a source of rough edges.
  • Chat-only OpenAI-compatible servers such as stock Ollama and llama.cpp do not support image generation.

Topics

#ai-agents#self-hosted#llm#personal-assistant#multi-agent-systems#telegram-bot#sqlite#chatbot#agents#ai-assistant

FAQ

What is hivekeep?
A self-hosted platform of persistent AI agents with long-term memory that collaborate and answer across chat channels, run in a single container. Hivekeep is a self-hosted platform for running a team of persistent AI agents on your own server. Each agent keeps one continuous session that does not reset, backed by a hybrid long-term memory that combines vector search with keyword search. Agents can message each other, delegate work to sub-agents, and run scheduled tasks.
Is hivekeep free to use? What license does it have?
hivekeep is open source under the MIT license. Check the license terms before using it in commercial products.
Is hivekeep still maintained?
Actively maintained. The last commit to hivekeep was on Oct 21, 2026.
What can I use hivekeep for?
Running a personal AI assistant that remembers context across months of conversations Operating a household or small team of specialist agents reachable from Telegram or Slack Keeping LLM API keys and account credentials on your own server Automating scheduled checks, reminders and email triage with agents
Where can I get hivekeep?
The source code and installation instructions are on GitHub: https://github.com/MarlBurroW/hivekeep

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects