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.
A well-structured, highly actionable workflow skill: every step has a concrete output template, BLOCKED gates enforce sequencing, and a verification checklist guards completion. Main gaps are the missing error-recovery feedback loop, an undefined $KEYWORD variable, and a mildly padded Anti-Rationalization section.
Suggestions
Add a feedback loop to the Verification Protocol: instruct what to do when a check fails (e.g. 'If any check fails, return to the corresponding step, fix, and re-verify before reporting completion') to reach anchor-5 workflow clarity.
Derive $KEYWORD explicitly (e.g. 'Extract 2-3 significant nouns from $ARGUMENTS and use them as $KEYWORD') and show how to loop the issue-breakdown table into gh issue create, since batch issue creation currently lacks validation.
Trim the Anti-Rationalization table to the 2-3 highest-value rows (or move it to a reference file) and drop the duplicated closing Output block to tighten token efficiency.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body assumes Claude's competence (no PRD/Gherkin/MoSCoW primers) and is dominated by tight templates and commands, e.g. "**BLOCKED** until at least 3 user stories with Gherkin criteria exist". Minor over-explanation that could be trimmed: the 7-row Anti-Rationalization table, the "bridging the gap between stakeholder intent and developer action" tagline, and partially overlapping codebase scans in Steps 1 and 4. | 4 / 5 |
Actionability | Mostly executable guidance: concrete bash ("grep -rl \"$KEYWORD\" src/ --include=*.ts...", "gh issue create --title ... --label \"feature,[size]\""), copy-paste output templates per step, and S/M/L/XL hour bands. Minor gaps keep it below a 5: $KEYWORD is never derived from $ARGUMENTS, and Step 7 shows one-off issue creation rather than how to iterate the breakdown table. | 4 / 5 |
Workflow Clarity | Seven clearly sequenced steps with explicit gates ("**BLOCKED** until problem statement is clear and measurable") and an 8-item Verification Protocol checklist — anchor 4 territory rather than 5 because there is no error-recovery feedback loop (nothing says what to do when a verification check fails) and the batch gh issue creation step has no pre-creation validation. Well above anchor 3, where checkpoints would be implicit rather than stated as BLOCKED gates. | 4 / 5 |
Progressive Disclosure | Self-contained skill with no bundle files; sections are well organized (When to Use, Workflow, Verification Protocol, Rules, Output) and the inline templates are used on every run, so their placement is appropriate. Not a 5 because at ~230 lines some always-loaded material (Anti-Rationalization table, per-step output templates) could be split into references to slim the core file. | 4 / 5 |
Total | 16 / 20 Passed |