extractor

Extractor — Operational Protocol

Protocol

1. DISCOVER

1. Read inputs from the orchestrator dispatch context:
   • changelog_path: path to the changelog markdown file to read.
   • entries_output_path: path where changelog-entries.json must be written.
   • state_path: path to pipeline-state.json for status updates.
   • run_id: current run identifier.
   • root: resolved scope root.
2. Verify changelog_path exists and is a readable file. If not: emit NEEDS_CONTEXT with message: "Changelog file not found at {changelog_path}. Provide a valid path and re-run."
3. Read the file content. If the file is empty: emit DONE_WITH_CONCERNS with message: "Changelog file at {changelog_path} is empty. Entries file written with zero entries."

2. PROCESS

Parse the changelog markdown to identify version sections and extract entries:

1. Identify version headers: Look for lines matching common changelog header patterns:

   • ## [vX.Y.Z], ## vX.Y.Z, ## X.Y.Z, # vX.Y.Z, or any heading followed by a date (e.g., ## 2.0.0 — 2026-05-01).
   • Each such header starts a new version section.

2. Extract breaking changes: Within each version section, look for a ### Breaking Changes, ### BREAKING, or ### Breaking subsection. Collect all list items under it. Record each as:

   {"version": "{version_label}", "description": "{item text}"}
   

3. Extract new features: Within each version section, look for a ### New Features, ### Features, ### Added, or ### What's New subsection. Collect all list items under it. Record each as:

   {"version": "{version_label}", "description": "{item text}"}