Codeteel

Open-source AI coding agent — Slack, Telegram, Discord, Web

8 followers

Open-source AI coding agent — Slack, Telegram, Discord, Web

8 followers

Self-hosted AI coding agent that connects to your local Ollama models. Code from Web, Slack, Telegram, or Discord. ReAct architecture, human-in-the-loop plan approval, deterministic edits, PR creation. Supports DeepSeek, Llama, Qwen, Mistral, OpenAI, Claude, Gemini. Free, open source, privacy-first.

Free

Launch tags:Developer Tools•GitHub•Tech

Launch Team

Viktor.comAn AI coworker that actually does the work

Promoted

Maker

📌

Hey Product Hunt! I'm Vicky, the builder behind Codeteel. I built this because every AI coding tool I tried had the same problem — my code was leaving my machine. As someone who works with enterprise clients, that's a non-starter. So I built an AI coding agent that runs entirely on your infrastructure. No code ever leaves your servers. Here's what makes Codeteel different: - Self-hosted — your code stays on your machine. Run local models with Ollama for zero API costs, or bring your own cloud provider (OpenAI, Claude, Gemini, Grok, and more) - Chat from anywhere — request code changes from Web, Slack, Telegram, or Discord. Whichever platform your team already uses - Not just autocomplete — it reads your entire codebase (vector embeddings for every file), understands context, generates a plan, and creates pull requests - Human-in-the-loop — every change goes through a review-and-approve flow before it touches your code. No yolo commits - Open source — fork it, extend it, audit it. 95%+ E2E test success rate across 30+ languages The workflow is simple: 1. Connect your GitHub repos → Codeteel indexes and understands your codebase 2. Ask for changes in chat → it generates a plan for you to review 3. Approve → it creates a PR with clean, deterministic edits Built with ReAct architecture, AES-256-GCM encryption, OAuth token security, and branch protection enforcement. I'd love to hear — what's stopping you from using AI coding tools at work today? Is it privacy, accuracy, or something else?

Report

3mo ago

Finally got it running against my local Ollama setup and the plan approval step genuinely surprised me, it's way more useful than I expected for catching bad ideas before they hit the repo.

Report

3mo ago

Ran it against a local Qwen model yesterday and the deterministic edits kept everything clean without me babysitting. Love that I can kick off a task from Slack without leaving the chat.

Report

3mo ago

Reviews

Maker

📌

Report

3mo ago

Finally got it running against my local Ollama setup and the plan approval step genuinely surprised me, it's way more useful than I expected for catching bad ideas before they hit the repo.

Report

3mo ago

Ran it against a local Qwen model yesterday and the deterministic edits kept everything clean without me babysitting. Love that I can kick off a task from Slack without leaving the chat.

Report

3mo ago