Skip to main content

​
Usage

archal harness list
archal run <scenario> --harness <name>
archal run <scenario> --harness <path>
archal harness scaffold [directory]
archal harness check <path>
archal harness check <path> --agent-model gemini-2.5-flash --json

​
Local harness paths

The preferred model is a real harness program inside the repo you want to test. Point --harness at an explicit local entrypoint or repo root: 
archal run scenario.md --harness ./.archal/harness.ts
If the repo does not have one yet, scaffold the standard entrypoint: 
archal harness scaffold .
Before the first real run, check that Archal can boot the harness headlessly: 
archal harness check ./.archal/harness.ts
You can also add --agent-model to match the model you plan to use and --json for machine-readable output. That command tells you whether the next missing layer is:
  • harness resolution
  • a runnable entrypoint
  • a headless boot issue
  • or a service bridge/runtime issue on the harness path
See Harness configuration for the runtime contract, environment variables, model tuning, and transport options.