Content
68%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 a well-structured, actionable, multi-step workflow with explicit confirmation checkpoints. Its main weakness is workflow clarity: the destructive archive move lacks a post-operation verification/feedback loop, which the rubric caps at 3.
Suggestions
Add a post-archive verification step after the mv command, e.g. confirm the source changeRoot no longer exists and the archive path now contains the moved directory, treating a failed move as an error to retry rather than silently proceeding.
Tighten the store-selection paragraph and the Output On Success template, which restate details already covered in the steps, to reduce token overhead.
Replace the 'YYYY-MM-DD-<change-name>' template in step 5 with a concrete command (e.g. date computation) so the archive naming is fully executable rather than left to inference.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is an efficient operational workflow built from concrete commands and JSON-field parsing without explaining concepts Claude already knows, with only minor padding in the dense store-selection paragraph and the restated output template, so it sits at 'efficient, minor trims possible' rather than the fully lean top anchor. | 4 / 5 |
Actionability | It provides mostly executable guidance with concrete commands (openspec list/status --json, mkdir -p, mv) and named tools (AskUserQuestion, Task with subagent_type), with minor gaps such as the date-stamped archive name left as a template ('YYYY-MM-DD-<change-name>') and sync delegated to a subagent prompt rather than a direct command. | 4 / 5 |
Workflow Clarity | Although the six steps are clearly sequenced with confirmation checkpoints for incomplete artifacts, tasks, and delta-spec sync, the archive step performs a destructive directory move with no post-move verification that the move succeeded and the source is gone, and the rubric caps destructive-operation workflows lacking validation at 3. | 3 / 5 |
Progressive Disclosure | The single self-contained SKILL.md is well-organized with clearly labeled sections (Store selection, Input, Steps, Output On Success, Guardrails) and only one-level references to the openspec-sync-specs skill, with minor organization gaps from the inlined store-selection detail and JSON-field enumeration rather than the cleanly-split structure of the top anchor. | 4 / 5 |
Total | 15 / 20 Passed |