What problem does it solve?
Rich-text editor projects like Slate and Plate need broad behavioral test coverage, but manually studying repositories like Lexical, ProseMirror, CodeMirror, or Tiptap for reusable test ideas is slow, inconsistent, and legally risky when licenses are unclear.
Core Features & Use Cases
- Exhaustive Multi-Pass Inventory: Builds a complete test-file inventory of a target editor repo, classifies every file (portable, plate-owned, skip, harness, product-shell), and extracts test names with line references.
- License-Aware Harvesting: Classifies the target repo as permissive or behavior-only, routes artifacts to durable docs or scratch directories, and enforces fresh-invariant rewriting instead of copying upstream test code.
- Coverage Mapping and Action Planning: Maps extracted behavior invariants to existing Slate v2 and Plate coverage, then assigns actions such as covered, refactor-existing, create-new, defer, or plate-owned with verification commands.
- Use Case: Point the skill at a local Lexical checkout to produce a scored harvest report identifying IME composition and selection-mapping tests that Slate v2 lacks, with concrete target files and verification commands for each gap.
Quick Start
Ask the agent to harvest the local ../lexical checkout for portable editor-behavior tests and produce a report with coverage mapping against Slate v2 and Plate.
