New Task
One clear instruction. ProPR files the issue, runs the agent, and opens the PR.
A task answers one request. A goal keeps a coding-agent session working toward an outcome across many steps. ProPR checkpoints the work on a draft pull request, shows you what the agent is doing, and takes your corrections while it runs.
ProPR has three ways to start work. They share the same repositories, agents, isolation and pull-request review.
One clear instruction. ProPR files the issue, runs the agent, and opens the PR.
A change you want to shape first. You approve the issue breakdown before any code is written. Planning
An outcome that takes many steps and judgment along the way. The agent works toward it while you watch and correct.
Choose how the agent turns the objective into commits.
ProPR opens a draft PR on the goal branch before the agent starts, and owns every commit and push. The agent requests checkpoints as it reaches stable points; ProPR validates the changed paths, commits, pushes, and posts visual previews to the draft PR. Checkpoints target a 15-minute cadence by default.
The agent decides the breakdown. It creates GitHub issues, starts and monitors their implementation through ProPR, and can assemble an epic PR from them. The goal finishes with a validated draft PR that holds the final implementation.
The goal console shows context, current activity, progress, artifacts, logs and visual previews. When the agent needs a decision, it asks.
Send an instruction or ask a question. The timeline records your words verbatim, next to what the agent did with them.
Stop at a checkpoint and pick up the same session later, or switch the model between steps. Work already committed stays on the branch.
Cap how many tasks an orchestrating goal runs in parallel, and choose whether Ultrafix polishes the result.
A goal ends in a draft pull request. You review, comment and merge in GitHub, as with every other ProPR change.
ProPR runs goals on each agent's own long-running session support, and offers an agent only when its configured image supports goals.
| Agent | Goal support | Corrections |
|---|---|---|
| Claude Code | Native | Live, while the agent works |
| Codex | Native | Live, while the agent works |
| Antigravity CLI | Resumable conversation | At the next safe boundary |
| OpenCode, Mistral Vibe | Tasks and plans only | Through PR follow-ups |
Through MCP, your chat client creates goals, reads their activity, and sends corrective input. Goals start from the web UI, the desktop app, MCP, or the propr goal CLI commands.
The Goals documentation covers options, checkpoints, recovery and limits.