kvyb

kvyb / opentulpa

Public

Self-hosted personal AI agent that lives in your DMs. Describe any workflow: triage Gmail, pull a Giphy feed, build a Slack bot, monitor markets. It writes the code, runs it, schedules it, and saves it. It also remembers everything you tell it and gets more useful over time.

25
2
100% credibility
Found Feb 23, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

OpenTulpa is a self-hosted personal AI agent accessible via Telegram that remembers user context, files, and preferences while autonomously building and scheduling custom tools and integrations from chat instructions.

How It Works

1
🔍 Discover your personal AI companion

You hear about OpenTulpa, a friendly AI that chats with you on Telegram and gets smarter over time by remembering what you share.

2
🤖 Create a Telegram bot

Chat with BotFather on Telegram to make your own bot and get a special link to start talking.

3
🔑 Connect smart thinking

Share a simple code from an AI service so your companion can understand and respond like a real friend.

4
🚀 Launch with one click

Run a quick start command on your computer, and your AI comes alive on Telegram instantly.

5
💬 Start chatting and sharing

Send messages, files, images, or voice notes—your companion remembers everything and uses it without reminding.

It learns and automates your life

Over time, it builds custom helpers, schedules reminders, and handles tasks just how you like, feeling like a personal assistant that truly knows you.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 25 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is opentulpa?

OpenTulpa is a Python-based, self-hosted AI agent that chats with you in Telegram DMs or via internal API, remembering everything you tell it—from preferences to files—for smarter responses over time. Describe any workflow like triaging Gmail, pulling a Giphy feed, building a Slack bot, or monitoring markets, and it writes the code, runs it in a sandbox, schedules recurrences, and saves reusable skills. Powered by FastAPI, LangGraph, and mem0 for memory, it runs locally with SQLite persistence and APScheduler for jobs, costing about $0.03 per task on OpenRouter models.

Why is it gaining traction?

Unlike cloud agents, it's fully self-hosted—no vendor lock-in or data leaks—acting like a self-hosted GitHub Copilot that builds and runs code from natural DM descriptions, including self-hosted GitHub Actions runners or bots in Docker. Developers dig the compounding intelligence: it gets more useful as you chat, auto-storing skills for feeds, automations, or personal finance trackers without context-switching to a dev environment. Low setup (two env vars, one script) and approval gates for safety hook tinkerers fast.

Who should use this?

Indie hackers automating personal finance or market alerts without SaaS fees. DevOps folks prototyping self-hosted GitHub Enterprise alternatives, like AI-driven runners for Codespaces or Actions. Power users tired of rigid bots, needing a DM agent to build custom Slack integrations or Giphy schedulers on their server.

Verdict

Promising early experiment for self-hosted agent fans—stars at 10 and 1.0% credibility signal raw prototype status with solid README and quickstart, but expect bugs and light tests. Try if you want a local code-building bot; skip for production until more polish.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.