Scenario 12: Drive HarnessTap from scripts or other agents
Use this when another automation layer needs structured output instead of human text.
Frequency: Occasional (often daily in CI, less often for interactive users) · Status: Shipped
Use this when another automation layer needs structured output instead of human text.
Typical commands:
harnesstap init --format json
harnesstap harness list --format json
harnesstap resource list --format json
harnesstap layer show my-setup --format json
harnesstap layer doctor my-setup --format json
harnesstap history . --format json
harnesstap layer apply my-setup --project . --dry-run --format json
harnesstap resource sync --format json --dry-runExit codes worth scripting against:
harnesstap layer doctorreturns exit code 1 when any doctor check reports an error severity finding.harnesstap status --checkreturns exit code 1 when drift exists.harnesstap layer apply --strict-plugin-versionsreturns exit code 2 when a pinned plugin's installed version violates its constraint.
This matters for agent-harness optimization because HarnessTap is not only a human CLI; it can also be the state and serialization layer that other agents inspect before deciding how to update a repo's assistant environment.
Scenario 11: Start from a catalog baseline instead of building from scratch
Use this when you want a fast starting point and only need to tailor it for your repo afterward.
Scenario 13: Choose a materialization strategy (symlink vs copy)
Use this when you need to decide whether a project's alias harness outputs are written as symlinks (atomic, always in sync with the main reference) or...