ProPR
Log in Get ProPR

Plan, run, review, and refine AI-generated pull requests.

ProPR covers every step that makes AI coding work in a real codebase: decide what changes, give the agent the right context, isolate the run, open the PR, and turn review feedback into the next commit.

Plan · Run · Review and refine · Operate

Gather context
Execute
Refine the PR

Feature overview

The public integry/propr pull request 1613 showing its implementation summary, execution record, model, duration, token usage, cost, and verification.
Public propr PR #1613 ties the approved issue, implementation, commit, and execution record together.

Plan

Decide what changes and show the model the code it needs.

A generated plan broken into five ordered tasks, each a future GitHub issue

Planning

Turn a prompt into a repository-aware plan and approve the scope before an agent writes a line.

Included Context (8 files): the courier-app files the planner will read, ranked by relevance, led by uploadQueue.ts and ProofOfDelivery.tsx

Context gathering

See every file, summary and screenshot the planning model will read, and fix the input before you spend a token.

The courier app's to-dos: a search box and the Offline mode category with its items

Repository todos

Capture follow-up ideas from anywhere, sort them by category and priority, and turn a related set into a plan draft.

Improvement suggestions for the courier app, two selected to turn into a plan or save as to-dos

Improvement suggestions

Scan the codebase for practical improvements, keep the ones worth building, and send them into todos or a plan.

Run

Every run is isolated, owned and recorded.

The end of an isolated run: the restricted network denied three hosts, the agent finished, and ProPR created the pull request

Isolated execution

Each task runs in its own Docker workspace, branch and Git worktree. The result arrives as a pull request.

Tasks with an Assignees column: each pull request shows the people GitHub has assigned, and the assignee filter narrows the list to me, to nobody, or to any teammates

Task assignment

Every task has an owner, synced with the GitHub assignee, so the Tasks screen shows what is yours.

An agent extension profile: the runbooks MCP server over HTTPS with two read-only tools and a write-only bearer token, plus the incident-runbooks skill pinned to a commit with its SHA-256 digest

Agent extensions

Give task runs your MCP servers and skills through extension profiles, tool by tool.

A paused goal with Resume and Cancel, and its draft PR: the same session picks up where it stopped

Goals

Give an agent a standing objective. It works in one long session, checkpoints into a draft PR and takes your corrections mid-run.

The definition of the weekly dependency review: its name, a description, and two repositories it may read

Automations

Save a prompt and run it on a schedule, by HTTP or from chat. Each run writes a Markdown report. Acting on the report is off until you turn it on.

Review and refine

The pull request is where the work gets judged, fixed and merged.

A review run on the pull request, scored 6 with two issues: it read the diff against the issue and checked the sticky summary on every checkout step

Review

A read-only critique on the PR: numbered merge blockers, optional suggestions and a score. You choose which fixes to apply.

One pull request’s runs: review scored 5, a fix, review scored 7, and the Ultrafix cycle in progress

Ultrafix

Name a score. ProPR alternates review and fix passes on the PR until it gets there.

A visual preview from the goal’s draft PR, loaded from GitHub into the goal console

Visual previews

The agent captures the screen it changed and ProPR posts the images or video on the pull request, next to the diff.

A /merge run on the pull request: merge main into the branch, resolve the conflict in CheckoutSummary.tsx, run the checkout tests

GitHub commands

A PR comment requests fixes, reviews, branch updates, model changes and reverts. The conversation drives every next step.

Operate

Run the same loop from chat, your phone or the desktop, and watch what it costs.

The MCP Log: a send_goal_input call succeeds, and a get_task call on a repository outside the grant is denied with FORBIDDEN_REPOSITORY

MCP from chat

Plan, start tasks and goals, read diffs and request reviews from the chat client you already use, within the repositories you grant.

Agent Tank in the sidebar: Claude session, weekly and Fable windows, the Codex weekly window, and Antigravity's weekly and five-hour limits for Gemini and for Claude and GPT

Agent Tank

See which tasks consume subscription capacity on Claude Code, Codex and Antigravity, and pause unattended runs near the limit.

Tasks on a phone: each pull request is a card with status, run track, agent and repository

Mobile

Push notifications and the Inbox tell you when a PR needs you. The GitHub mobile app is the remote control.

ProPR Desktop connecting to an instance: a display name, Northwind Labs, the instance URL and a Connect button

Desktop app

A client for your instance on Linux and macOS, with a keyboard shortcut for every menu action.