Open-source, cross-platform AI chat client for Web, desktop and mobile that connects to OpenAI, Claude, Gemini, DeepSeek and self-hosted models.
nalgeon/pokitoki
A self-hosted Telegram chat bot that answers questions using OpenAI, Claude, Gemini, or any OpenAI-compatible model.
- Stars
- 341
- Forks
- 60
- Open issues
- 0
- Last commit
- 14 days ago
What is it?
Humble AI Telegram Bot is a chat assistant that runs inside Telegram and forwards questions to large language models from OpenAI, Claude, Gemini, or any OpenAI-compatible provider such as OpenRouter. It works in one-on-one chats and in groups, where it replies to direct mentions or to replies on questions.
The bot keeps conversation context for a configurable timeout, supports per-chat custom prompts and models, and lets admins change most settings on the fly with the /config command. Built-in shortcuts such as proofread, translate, and summarize handle common writing and code tasks, and custom shortcuts can be added in the config file.
With the OpenAI, Claude, and Gemini providers, the model can search the web and open links using server-side tools. Other setups fetch linked text content for the model. Groups can be given per-user message limits to control API spending, and the bot is deployed with Docker Compose.
Key features
- 01Works in personal chats and groups, answering direct questions and mentions.
- 02Supports OpenAI, Claude, Gemini, and any OpenAI-compatible provider.
- 03Keeps conversation context and continues follow-up questions within a set timeout.
- 04Lets each chat use its own prompt and model.
- 05Provides shortcuts like proofread, translate, and summarize, plus custom ones.
- 06Changes most settings at runtime through the /config command.
Use cases
- Running a private AI assistant for yourself or a small group of friends on Telegram.
- Giving a team a shared AI helper in a Telegram group with usage limits.
- Proofreading, translating, or summarizing text directly from a chat.
- Asking questions about text files or linked web pages inside Telegram.
Quick start
- 1
Get an AI API key and a Telegram bot token from BotFather
- 2
Clone the repository
git clone https://github.com/nalgeon/pokitoki.git - 3
Copy the example config to config.yml and add your tokens
cp config.example.yml config.yml - 4
Start the bot with Docker Compose
docker compose up --build --detach
Things to know
- You must supply your own AI provider API key and Telegram bot token, and provider usage may be billed.
- Web search and link tools may incur extra charges from the provider.
- The project only accepts bug fix contributions, not new features.
In collections
Topics
FAQ
- What is pokitoki?
- A self-hosted Telegram chat bot that answers questions using OpenAI, Claude, Gemini, or any OpenAI-compatible model. Humble AI Telegram Bot is a chat assistant that runs inside Telegram and forwards questions to large language models from OpenAI, Claude, Gemini, or any OpenAI-compatible provider such as OpenRouter. It works in one-on-one chats and in groups, where it replies to direct mentions or to replies on questions.
- Is pokitoki free to use? What license does it have?
- pokitoki is open source under the MIT license. Check the license terms before using it in commercial products.
- Is pokitoki still maintained?
- Actively maintained. The last commit to pokitoki was on Oct 24, 2026.
- What can I use pokitoki for?
- Running a private AI assistant for yourself or a small group of friends on Telegram. Giving a team a shared AI helper in a Telegram group with usage limits. Proofreading, translating, or summarizing text directly from a chat. Asking questions about text files or linked web pages inside Telegram.
- How do I install pokitoki?
- Get an AI API key and a Telegram bot token from BotFather Clone the repository: git clone https://github.com/nalgeon/pokitoki.git Copy the example config to config.yml and add your tokens: cp config.example.yml config.yml Start the bot with Docker Compose: docker compose up --build --detach
- Where can I get pokitoki?
- The source code and installation instructions are on GitHub: https://github.com/nalgeon/pokitoki
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Self-hosted AI gateway that puts API keys and subscription accounts from many providers behind one endpoint, with scheduling, failover and usage logs.
A Go chat bot that connects Telegram, Discord, Slack, Lark, DingTalk, WeCom, QQ and WeChat to LLMs such as OpenAI, Gemini and DeepSeek.
Self-hosted admin system for integrating many LLM providers (OpenAI, Claude, Gemini, DeepSeek, Qwen and others) behind one unified API.
Self-hosted ChatGPT-style chat platform that unifies many AI providers, with Agents, MCP support, Artifacts, and multi-user auth.
A self-hosted AI agent platform where teams share agents, tools, skills and memory across channels like Telegram, WhatsApp, Discord and Slack.