marrouchi/slack2pr

Slack2PR lets you mention a bot in Slack, describe a feature or a bug in plain English, and have it plan the work, write the code, test it, and open a pull request on GitHub, replying in the same Slack thread with the link. It is built as an app for a chatbot platform called Hexabot. When you mention the bot, it first figures out what you want: build a new feature, investigate and fix a bug, or just answer a question about the code. For a new feature it asks a short series of questions to understand the requirements, breaks the work into smaller pieces, and builds each piece one at a time. For a bug report it first looks at the code without changing anything, then lets you choose to dig deeper, fix it, or stop. For a question it just reads the code and replies with an answer. The actual coding work happens inside a temporary Docker container that holds a copy of your repository. Each Slack conversation gets its own container, which is reused across the whole conversation so the plan and progress are not lost between steps, and is removed once the conversation ends or goes quiet. You can choose which underlying AI coding tool does the work, including Claude Code, Codex, OpenCode, or Grok Build, and pick the model for each step. Git and GitHub access are set up automatically so the AI can create branches, commit, push, and open pull requests without ever seeing your raw access token directly. Built in rules block risky commands like sudo or deleting files, and control what files can be read or written. To run it you need Node.js, Docker, a Slack app connected to your workspace, a GitHub access token, and an API key for an AI model provider. After starting the app you create these credentials in an admin screen, import the included workflow, point it at your repository, and connect it to Slack. The README does not state a license.