Content
71%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is exceptionally concise and well-structured as a rules sheet, but it functions as invariants rather than an actionable workflow, offering no concrete code, sequencing, or validation checkpoints for the destructive/migration-sensitive operations it touches.
Suggestions
Add a short worked example or command for read-your-writes (e.g., the specific projector-await call or check) so the rule is executable, not just a directive.
Include an explicit validation checkpoint for compatibility-affecting changes (e.g., 'before changing an aggregate lane, run the replay-compatibility check and confirm the projection catches up').
Consider offloading the deeper material on OCC, projections, and migration safety to a one-level-deep reference file (e.g., PROJECTIONS.md) and linking it from the relevant rules.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | A lean five-bullet 'Core Rules' section with no concept padding; every line states a durable rule Claude would not already infer, so every token earns its place. | 5 / 5 |
Actionability | The rules are clear directives but provide no concrete code, commands, or executable examples — e.g., how to actually wait for a projector, what 'replay compatibility' check to run, or how migration safety is verified — leaving guidance high-level rather than copy-paste ready. | 3 / 5 |
Workflow Clarity | There is no sequenced workflow or validation checkpoint; the rules are atomic invariants. Operations implied here (writes needing read-your-writes, changing an aggregate lane) are risky/batch-like and lack explicit validate-then-proceed steps, so the cap-at-3 guidance for missing validation applies. | 3 / 5 |
Progressive Disclosure | Well-organized under a single 'Core Rules' heading with no nested references; for a short skill this is good structure. It does not reach 5 because the body points to no one-level-deep reference files for the larger concepts it names (projections, OCC, migration safety), which could offload detail. | 4 / 5 |
Total | 15 / 20 Passed |