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.
A well-structured, highly actionable instruction-only skill with a clear 7-step workflow, explicit validation checkpoints, a completion checklist, and well-signaled one-level references. Its main weaknesses are dead links to nonexistent shared/base policy files, an unreferenced bundled validation script, and some redundancy across activation/routing sections.
Suggestions
Remove or fix the links to shared/base/*.md policy modules and shared/workflow-contract.md, which do not exist in the bundle — either ship those files or drop the framework boilerplate section.
Reference scripts/validate_scenarios.py from the body (e.g., in the Output contract or Completion checklist) so the bundled validator is discoverable and the output contract is actually checked.
Consolidate the overlapping trigger and routing content in "Purpose and activation" and "Routing Boundary" into a single section to save tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and directive ("Use `timestamptz` for instants; store money as integer minor units plus currency"; "expand -> migrate/backfill -> verify -> switch -> contract") and assumes Claude's competence without teaching basics, but the Base Framework boilerplate, the duplication between "Purpose and activation" and "Routing Boundary", and the lengthy workflow-integration/state.json mechanics are trimmable, matching the 4-anchor's minor over-explanation rather than the 3-anchor's unnecessary explanation. | 4 / 5 |
Actionability | Guidance is concrete and executable for an instruction-only skill: an exact output-contract heading list, a per-table mini-specification template, specific commands like `EXPLAIN (ANALYZE, BUFFERS)`, and named tooling. It falls short of the 5-anchor because the Examples section points to references rather than demonstrating a worked case, and the bundled validate_scenarios.py script is never invoked. | 4 / 5 |
Workflow Clarity | A clearly sequenced 7-step workflow with explicit validation checkpoints ("Validate candidates using representative `EXPLAIN (ANALYZE, BUFFERS)` after data exists", migration "verify" step, "abort criteria, and tested rollback or forward-repair") and a completion checklist, so the destructive/batch cap does not apply. It sits between the 4- and 5-anchors because error recovery is stated as an output requirement rather than an explicit validate-fix-retry step in the workflow, and the shipped validation script is orphaned. | 4 / 5 |
Progressive Disclosure | The four references/*.md files are real, one level deep, and clearly signaled ("Read [architecture-and-modeling.md](references/architecture-and-modeling.md)"), but scored against the actual bundle: roughly ten links to shared/base/*.md policies and shared/workflow-contract.md point to files that do not exist in the bundle, and scripts/validate_scenarios.py is never referenced from the body. These dead links and the orphaned script exceed the 4-anchor's "minor organization gaps". | 3 / 5 |
Total | 15 / 20 Passed |