Harness Intelligence Wiki
HarnessEntry Points

Entry Points

Choose the right Harness starting point for workflow, atomic skill, or CLI-driven work

Harness entrypoints are the supported ways to start useful work with Harness Intelligence. Use this page when you know the kind of help you need and want the shortest route into the right workflow, skill, or command capability.

Get Started remains the first-run path for installation and first command guidance. This overview answers what Harness can do once you are choosing the work mode.

Core

Core entrypoints are the major workflow starts. They turn a project intent, accepted scope, runtime symptom, branch, finding set, or docs change into a bounded Harness run.

AFK delivery is an implementation child route for unattended delivery: AFK Implementation.

Atomic

Atomic entrypoints are intentionally small operating moves. They can start useful Harness work without a full lifecycle wrapper.

  • parallel-research: split independent readonly investigation across bounded questions, paths, or hypotheses, then converge on evidence.
  • handoff: preserve the useful continuation state when context should move across agents, sessions, or time.

Commands

Command capabilities start from the dp CLI. The overview names the capability; the CLI docs own mechanics.

  • dp update: refresh generated Harness scaffold files and inspect scaffold drift.
  • dp report: create a report with the evidence needed for triage and follow-up.

Where To Go Next

Choose Core when the work needs a durable workflow. Choose Atomic when the next move is one skill-sized operating action. Choose Commands when the task starts from local scaffold state or a reportable issue.

On this page