Skip to main content

Orchestrator

resume-pipeline-orchestrator is the execution layer that ties the skills together for live JD processing.

For the full operator path, see Resume Agent > Run and Verify. This page stays focused on the orchestrator's role inside the broader pipeline.

Responsibilities​

  • fetch unprocessed job descriptions from the dashboard
  • load candidate-profile
  • skip JDs that are disqualified, fail the binary gate, or score below 40
  • run jd-extraction, project-selection, point-repointing, and latex-assembly sequentially for JDs scoring 40+
  • save the generated .tex locally
  • run the mandatory self-review gate before any push
  • push successful resumes, PATCH /use, and log outcomes

Control flow​

This is the control flow the orchestrator owns during a live JD-processing run.

  • The orchestrator is the runtime entry point.
  • Users should not manually chain downstream stages in normal operation.
  • Success and failure paths both end in logging.

Important rule​

The orchestrator should consume candidate-specific truth from candidate-profile. It should not introduce its own hidden assumptions about location, sponsorship, seniority, or role fit.

Manual one-off debugging can invoke individual skills directly, but normal JD processing should start with resume-pipeline-orchestrator only.

If you want the actual orchestrator and downstream skill files referenced here, use the public source repository.

hermes-autonomous-resume on GitHub