Content
75%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 delivers a tight, executable workflow: concrete commands, an exact template, an explicit user gate, and failure-path notes. Remaining polish is minor — undefined terminology ("ledger"), one malformed heading, redundant emphasis of the field-name constraint, and no self-validation of the written handoff.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is nearly all project-specific instruction (exact paths, an exact YAML template, copy-paste commands) with no explanation of concepts Claude already knows; minor trimming is possible since the "use exactly goal:/now:" constraint is stated three times ("CRITICAL: Use EXACTLY this YAML format", "they MUST be named exactly this", "The statusline parser looks for EXACTLY goal: and now:") and the Additional Notes partially restate the intro. | 4 / 5 |
Actionability | Every step is backed by executable material: copy-paste bash for deriving the folder name and for indexing/marking, an exact YAML template, exact AskUserQuestion options, and a final response template. Minor gaps remain — "session: {session-name from ledger}" references an undefined "ledger" (step 1 derives the name from handoff folders), and the opc/scripts/ commands assume an unexplained project layout. | 4 / 5 |
Workflow Clarity | Four clearly numbered steps with a user-response gate before marking and a note explaining the failure path ("Database marking was not available"); it is not 5 because there is no validation of the written YAML itself and no recovery loop if artifact_mark.py fails. The destructive/batch cap does not apply since this is a single-file write plus index, not a batch or destructive operation. | 4 / 5 |
Progressive Disclosure | A well-sectioned single-file skill with no bundle files; the YAML template and commands genuinely belong inline and nothing warrants splitting into references. It is not 5 because at ~137 lines it exceeds the simple-skill threshold, the final heading is malformed ("##."), and repeated warning blocks could be consolidated for easier navigation. | 4 / 5 |
Total | 16 / 20 Passed |