Assign a task. AI opens the PR.
Point an AI coding agent at any task in your plan. It reads the full brief context, understands your codebase, writes the code, and opens a pull request — while you focus on something else.
How it works
Brief context all the way to the PR.
The agent doesn't just get the task description. It gets the full brief — goals, requirements, research, constraints. It understands not just what to build, but why. That context leads to implementation decisions that actually make sense.
Select a task
Navigate to any task in a brief's plan. The task title, description, acceptance criteria, and subtasks are all passed to the agent.
Configure the run
Specify the repository URL and branch. Add optional context instructions: "follow patterns in the auth module" or "avoid changing the schema".
Launch and leave
The agent starts immediately. Close the tab, work on something else. Status updates arrive in real time.
Review the PR
When the agent finishes, it opens a pull request. The run record in Hamster links directly to the PR.
Run states
Pending
Agent is starting up
Running
Actively working in the repository
Completed
Finished and PR is open
Failed
Error occurred; run details available
Key capabilities
Full brief context
The agent reads the brief document alongside the task — it understands the project before writing any code.
Non-blocking
Close the tab, start another session. The agent runs independently in the background.
Pull request creation
By default, the agent opens a PR when it finishes. The PR is linked back to the task run in Hamster.
Real-time status sync
Webhook-based updates keep the task view current as the agent progresses.
Run history
Every run is recorded with status, timestamps, and outcome. Review the history of runs on any task.
Additional context
Provide free-text instructions before launching: constraints, patterns to follow, things to avoid.
Let the agent run.
Free to start. No credit card required.