Coding-agent work, independently verified
Agentic Harness
No setup required
See a verified result in about a minute
Run the real harness on a temporary practice project with a scripted worker. It uses no AI model or API key, and it never touches the project shown above.
Practice run — this demonstrates the workflow and evidence, not model quality.
New task
What would you like done?
Describe the result in your own words. The assistant will plan, work, and show evidence before calling it done.
A normal sentence is enough.
Choose how much time the assistant should spend. This does not change where the work runs.
Standard plans the work and verifies the result.
Before you start
What to expect
Loading the execution plan…
- Runs
- Checking…
- Model
- Automatic
- Changes
- Implementation
View route details
- Planner
- Automatic
- Executor
- Automatic
- Verification
- Independent checks
- Maturity
- Supported
Change how it runs
Choose where the task runs and which systems plan, execute, and review it.
Automatic follows this installation's recommended model. Choose a named profile only when the task needs it.
Labs · bounded experiments
These routes are limited trials. Availability and scope requirements come from the connected backend.
Checks · Automatic
Agentic Harness runs this independently. Technical commands remain available here for advanced users.
Access · Entire project
Leave this blank to allow the whole project, or limit the task to specific project folders.
Describe the outcome you want. The app will explain anything else it needs.
Task history
Previous tasks
Review completed work and the evidence saved with each task.