Every provider runs through one path
Claude Code, Codex, Antigravity CLI, Mistral Vibe and OpenCode, on subscriptions or API keys, all ship through the same loop.
Adopt AI coding inside a disciplined engineering process.
The hard part of adopting coding agents is governing what they ship. ProPR gives the whole team one process for planning, pull requests and review, with a full history on every agent change.
What every task carries
- Cost visibilityPer-call model, tokens, cost, and duration in the usage log.
- Access controlYou decide who can trigger tasks and comment commands.
- Traceable logsPrompts, plans, usage, commits, and follow-up history on every task.
- Provider choiceFive agents, subscriptions or API keys, one process.
- One repeatable pathEvery agent and developer ships through the same plan, PR, and review loop.
Overview
Agent work in private terminal sessions leaves one-off branches, unclear spend and pull requests nobody can explain. ProPR puts every planned change on one path with shared context, review, follow-up commits and an audit trail. Developers keep their own tools for exploration.
One visible path makes AI changes governable.
Every task walks the same loop, whichever agent runs it and whoever starts it: plan, isolated run, pull request, review, refine, merge.
01
Plan
ControlAn allowed user starts the task and approves the scope before code changes begin.
RecordPrompt, context, task split, chosen agent and model.
02
Isolated run
ControlThe agent edits inside its own Docker workspace, Git worktree, and branch.
RecordCommits, logs, model, tokens, cost, duration.
03
Pull request
ControlEvery result arrives in GitHub as a branch and a pull request.
RecordDiff, implementation summary, execution record.
04
Review
ControlReviewers comment, CI reports, and
/review posts a scored assessment on request.
RecordReview findings, scores, comment history.
05
Refine
ControlPlain PR comments become follow-up commits on the same branch;
/fix applies pending findings.
RecordFollow-up commits, each explained in a comment.
06
Merge
ControlA person merges the pull request unless you enable auto-merge for green PRs.
RecordNormal GitHub merge history on the base branch.
Access control spans the whole pathA user whitelist decides who can trigger tasks from the dashboard, CLI, and issue labels. ProPR ignores bot accounts and runs comment commands only for allowed authors.
Cost shows up per task
The usage log ties each call's model, cost and duration to the change that used it. The cost comparison weighs subscriptions, API usage and seats together.
Agents get a read-only token and your network rules
Credentials, task history, logs and workspaces live in your deployment. Agents hold a read-only GitHub token, and optional restricted network mode limits them to an allowlist of hosts. Trust and security lays out the boundary.
Recovery is part of the path
Retry, discard, revert or correct a wrong attempt from the PR conversation.
What each role approves.
The administrator approves access, models and network rules
Who can trigger tasks, which provider runs each phase, and what a run can reach on the network.
The author approves the plan
Whoever starts the task approves the scope before any code changes. Planning shows the context the model will read and splits large changes into review-sized pieces.
The reviewer approves the pull request
Reviewers comment, CI reports, and /review adds a scored assessment. The assignment gate limits follow-ups to a PR's assignees. A person merges, or auto-merge handles green PRs.
Good first use cases.
Start where the boundaries are visible: small features, contained bug fixes, refactors with tests, repository cleanup, documentation tied to code, and PR review follow-ups.
What to measure.
How many planned tasks become merged PRs, how many follow-ups each needs, what each task costs, and which agents earn each phase. Dashboards and analytics report all four.
How to roll out.
Connect one repository
Follow the quickstart, choose who may trigger tasks, and pick the agents and models for each phase.
Ship the first use cases through pull requests
Move planned features, fixes, refactors and PR follow-ups into ProPR. Keep direct agents for live exploration.
Widen the loop
Add repositories, turn on comment commands and the assignment gate, and let Ultrafix or an epic chain carry larger changes.
Review the record
Use the usage log and task history to decide which models earn each phase and where routine changes run with less supervision.