ProPR
Log in Get ProPR

AI coding agents, governed the way your organization already ships.

ProPR runs Claude Code, Codex, Antigravity CLI, Mistral Vibe and OpenCode through one planned, isolated, reviewed pull-request workflow, on infrastructure you control. Every change carries its plan, logs, model and cost, and your team merges it in GitHub.

Book a demo Read the data boundary →

What the enterprise gets

  • Your infrastructure or a dedicated instanceSelf-host the Apache 2.0 stack, or let the ProPR team run a VM that is yours alone.
  • Isolation on every runOwn worktree, branch and Docker workspace; a read-only GitHub token in task containers, with write access kept only by orchestrating goals; optional restricted egress.
  • An audit trail in Git and GitHubPlan, prompt, logs, model, tokens and cost on every task, plus the PR record itself.
  • Enterprise plan and supportSSO/SAML, RBAC, org-wide telemetry, a dedicated Slack channel and an SLA.

Operational overview

Deploy it your way.

Same open-source ProPR, same agents, same workflow. You choose who operates the machine.

Self-hosted: everything on your infrastructure.

Published Docker images run the full stack on a Linux host inside your network. Repositories, credentials, logs and task history are stored and operated on that stack, and the Apache 2.0 source is yours to audit. The prompt and code context for each task go to the AI provider you select for it, and branches, commits and pull requests go to GitHub. Optional features add recipients: the Connect relay carries full GitHub webhook payloads, Connect Plus keeps preview originals when you opt in, and MCP returns results to your chat host and its model service. The trust page lists every recipient and what it holds.

  • Free, with no developer limit.
  • Your own GitHub App, or the Connect relay with no inbound endpoint.
  • A documented path to a hardened server.
Self-hosting →

Dedicated Hosted: a VM that is yours alone.

The ProPR team provisions, upgrades, backs up and restores a dedicated instance for you. Connect routing, the managed tunnel and encrypted backups are included; AI usage still goes to your own provider.

  • Plans from one developer to 15 seats and 12 concurrent runs, with extra seats on top.
  • Encrypted application backups, kept 60 days.
  • No public inbound ports on your VM.
ProPR Hosted →

A data boundary you can draw on a whiteboard.

Every external system has one role. Each receives only what that role requires, and the rest stays in your deployment.

GitHub hosts review

Branches, pull requests, review comments, status updates and inline preview copies. Plans, task history, logs and usage records stay in ProPR.

Your provider sees one task

The agent you select for a plan, implementation, review or follow-up receives the prompt and the code context for that task. You pick the provider per phase.

Your infrastructure runs the rest

Docker workloads, the database, queue, repository clones, Git worktrees, indexes and the audit trail. Who gets what, in full.

Isolation and an audit trail on every run.

Each run gets its own workspace

Every implementation gets its own Git worktree, branch and Docker workspace. The agent holds a read-only GitHub token, so ProPR makes every branch, commit, push and pull request. Isolated execution.

Restricted network egress, enforced when you say so

Optional restricted mode starts implementation, follow-up, fix, review, goal and indexing containers with no network and routes their traffic through a per-run allowlist proxy, with no elevated privileges. Set it per instance, enforce it for every repository, and read every denied host on the task timeline. Plan generation stays on the open network, and Antigravity falls back to open mode, or is refused when enforcement is on.

Every task keeps its record

The web UI and CLI show the plan, prompt, logs, model, tokens, cost and duration of every run, and the usage log ties each model call to the change that used it. Operations and analytics.

The history lives in tools you already trust

Every change lands as Git objects and GitHub records: branch, commits, PR conversation and review scores. Clone, grep and diff the history with nothing new to learn.

Access controls that ship today.

Admin and member roles

Roles map to GitHub accounts, with an audit trail of every change. Administrators decide which agents run by default and how much automation a repository allows.

Who may trigger work

Choose who can start tasks, and which labels and comments count. The assignment gate limits follow-ups on a pull request to its assignees.

A scope ceiling for chat

An administrator switches MCP on and sets the ceiling. Each user grants repositories on a consent screen, revokes apps in Connected apps, and the MCP Log records every tool call and its outcome.

Human merge, or auto-merge on green

A person merges in GitHub. Where you allow it, green pull requests merge themselves, and you decide which checks may block that per repository.

The Enterprise plan.

Organization-wide identity, access and visibility, with the ProPR team on call.

SSO/SAML

Sign in through your identity provider.

Role-based access control

Permissions by role across the organization.

Org-wide telemetry dashboards

One view across every repository in the organization.

Dedicated Slack channel

The ProPR core team one message away.

SLA

A committed response time, in writing.

Scoped to your organization

Tell us your team size, repositories and deployment, and you get a plan and a price.

Book a demo

Priority support and a founder on the call.

Enterprise Support Retainer

The ProPR core team one message away, priced by engineering team size.

  • Private Slack channel for up to 3 technical leads
  • 24-hour SLA on workflow interruptions
  • Monthly 60-minute strategy call
Request a quote →

Deployment and onboarding

Book a founder for the rollout itself.

  • Live Deployment Pairing. A 2-hour remote session that deploys ProPR into your AWS, GCP or Linux VPC, keys and webhooks configured.
  • Custom CI/CD Integration. Fixed-scope engineering that connects ProPR to your internal tools, Jira and Linear transitions and legacy test frameworks.
  • Team Onboarding. A 3-hour session for up to 30 engineers, on your projects, recorded for everyone who joins later.
Services and prices →

Questions enterprise buyers ask.

Does ProPR hold compliance certifications?

No. ProPR holds no compliance certifications. The Apache 2.0 source, the trust page and the privacy policy document exactly what runs where, and self-hosted ProPR stores code, credentials and agent activity on your own infrastructure. Each task's prompt and code context go to the AI provider you selected, and ProPR writes branches and pull requests to GitHub. The optional Connect relay, Connect Plus preview storage and MCP add further recipients, each listed on the trust page.

Do agents get write access to our repositories?

Only orchestrating goals do. Implementation, follow-up, fix, review and direct-goal containers hold a read-only GitHub token, and ProPR itself creates every branch, commit, push and pull request. An orchestrating goal keeps write access to drive the issues it files, and your team still reviews and merges in GitHub.

Who pays for AI usage?

You pay your providers directly, through the subscriptions you already hold or metered API billing. ProPR adds no token markup and shows the model, tokens and cost behind every pull request.

Can we keep agents off the open internet?

Yes, for implementation, follow-up, fix, review, goal and indexing runs. Restricted network mode starts those containers with no network and routes their traffic through a per-run allowlist proxy on the worker: the agent's provider API, GitHub, the npm and PyPI registries and the hosts you add. Enforce it for the whole instance, and read every denied host on the task timeline. Plan generation stays on the open network, and Antigravity falls back to open mode, or is refused when enforcement is on.

Which agents can we standardize on?

Claude Code, Codex, Antigravity CLI, Mistral Vibe and OpenCode, chosen per repository and per phase. Administrators decide which agents and models are allowed.

See ProPR on your repositories.

A walkthrough with the founder: your deployment options, the data boundary, and the first pull request your team would ship.

Book a demo