Analysis updated 2026-05-18
Delegate a feature request to an AI pipeline that plans, codes, tests, reviews, and opens a PR
Direct code changes to your repo from a Telegram chat instead of a terminal
Run isolated AI coding workspaces for multiple projects or team members at once
Send a voice message describing a feature and have it transcribed into a coding task
| duckbugio/flock | ninehills/pdf2md | internetkafe/cfsearch | |
|---|---|---|---|
| Stars | 36 | 36 | 35 |
| Language | Go | Go | Go |
| Setup difficulty | hard | hard | moderate |
| Complexity | 4/5 | 4/5 | 2/5 |
| Audience | developer | developer | ops devops |
Figures from each repo's GitHub metadata at analysis time.
Requires a Claude subscription or Anthropic API key plus a Telegram bot and Docker.
DuckFlock is a service you run on your own server that turns Claude, Anthropic's AI system, into a small software development team you can direct through Telegram. You describe what you want built in a Telegram chat message, and a pipeline of automated AI roles takes it from there: one role plans the work, another writes the code, a third runs tests, a fourth reviews the changes and posts comments, and a fifth decides whether the result is ready or needs another pass. When the work passes review, the system opens a pull request in your code repository automatically. The setup runs as a Docker container, which is a self-contained package you can start with a few terminal commands. You connect it to your Telegram bot account, authorize which Telegram users are allowed to send it instructions, and optionally link it to a Git host such as GitHub, Gitea, or GitLab so it can clone repositories and create branches. Each Telegram conversation gets its own isolated workspace on disk, so different chats cannot see each other's files, and multiple conversations can run at the same time. The system works with a Claude subscription (the same account type used in Claude.ai) or an Anthropic API key, which controls how the AI usage is billed. Voice messages are also supported: record a message in Telegram describing what you want, and the system transcribes it and treats it as a written instruction. The project is primarily aimed at developers who want to delegate routine feature work or code changes to an AI pipeline without manually prompting each step. It is written in Go, MIT licensed, and structured so that Telegram is just one adapter on top of a shared core, with other chat platforms expected to be added later.
A self-hosted service that turns Claude into a small AI dev team you direct over Telegram, from planning through opening a pull request.
Mainly Go. The stack also includes Go, Docker, Telegram Bot API.
Free to use, modify, and distribute for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.