Run ProPR where your code and credentials already are.

ProPR's Apache 2.0 code and published Docker images run on your infrastructure. Your repositories, credentials, CLI, agents, logs, and task history live in the stack you run. The quickstart gets it going with two commands.

Your Docker host Self-hosted

GitHubevents, branches, PRs

events ↓ ↑ git

Selected providerdirect model calls

Published ProPR servicesone pinned image set

Web UI + API Queue + workers Database + repos

Task worktree Abranch + agent container

Task worktree Bparallel and isolated

Mounted stack directoryconfiguration, database, logs, repository clones, and credentials

The CLI or launcher starts sibling services on your Docker host. Task worktrees and agent containers remain inside that deployment boundary.

A Docker stack you control.

A request becomes a plan, the selected agent runs in isolation, and the result returns as a pull request. The prebuilt stack runs on Linux amd64 with Docker and keeps its state locally. Guided propr setup connects GitHub through ProPR Connect; advanced deployments bring their own GitHub App. Agent Tank adds optional capacity tracking. System requirements.

The CLI needs Node.js 22+. propr setup runs host checks, agent sign-in, GitHub, and startup in one pass; the individual commands run each step on its own. Without Node.js, the propr/launcher image does the same job.

npm install -g propr-cli
propr setup

propr init stack
propr check
propr start

A thin wrapper around a stack you own.

What does the CLI do?

The propr CLI checks the agent credentials on your host, scaffolds the stack root, and starts the published images on your Docker daemon. Every generated file sits in the stack directory, ready to inspect, diff, or edit. The documentation covers every path and port.

Happening now: five runs in parallel, among them an implementation on storefront-web and a review on orders-api, each with its step and elapsed time — Your own instance at work: each running agent with its repository, step and elapsed time.

Source, images, and intake.

Source repository

The full ProPR codebase under Apache 2.0. Read it, self-host it, adapt it, contribute back.

Docker images

Docker Hub carries the app, web UI, docs, launcher, and one unified agent image with every supported agent: Claude Code, Codex, Antigravity, OpenCode, and Mistral Vibe. The launcher pulls and runs the stack with one command.

Event intake

Issues and PR events arrive over the hosted WebSocket relay by default, so you need no inbound endpoint; webhooks and polling work too. The documentation covers credentials, directories, updates, and a hardened production install.

Control is the point of self-hosting.

ProPR handles plans, prompts, repositories, task logs, provider credentials, and GitHub actions, so where it runs matters. Trust and security lists what stays in your deployment and what GitHub and each model provider receive. The guide to running AI coding agents on your own server adds the host it takes.

Repository state lives in your stack

Repository clones, task files, indexing state, logs, and worker state live in the ProPR stack you run.

You pick every provider

Choose Claude Code, Codex, Antigravity CLI, Mistral Vibe, or OpenCode, on subscriptions or direct APIs, per repository and per task.

Review happens in GitHub

Branches, PRs, comments, checks, and merges happen in GitHub.

Connect chat directly to your instance.

The optional direct MCP route serves https://<instance>/api/mcp with your instance's own OAuth and bypasses Connect. Agents and execution run on your instance; results you request reach your chat host. The hosted route through Connect's Plus managed tunnel reaches the same endpoint. MCP setup guide.