Content
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A thorough, highly actionable migration workflow with strong validation feedback loops and well-structured progressive disclosure into real reference files. The only weakness is mild verbosity from cross-step repetition that could be consolidated.
Suggestions
Consolidate the partial-file-inclusion (:start-after:/:end-before:) guidance so it lives in one place (Step 7 or Edge Cases) rather than being restated, to reduce token cost.
Centralize the sample-data teardown rule ('do NOT drop sample databases') once and reference it, instead of repeating it in Steps 6, 8, and 10.
Consider moving the large Step 12 report-template/example tables into a reference file to keep the main body leaner while preserving the copy-paste templates.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and mostly free of concept-explanation padding, but its length and repetition (partial-file inclusion appears in both Step 7 and Edge Cases; drop-database cautions recur across Steps 6/8/10) mean it could be tightened without losing clarity. | 2 / 3 |
Actionability | Provides concrete, copy-paste-ready guidance throughout: specific Bluehawk tags (:remove:/:replace:/:uncomment:), decision tables, a full reviewer subagent prompt template, and example report/discrepancy tables. | 3 / 3 |
Workflow Clarity | Steps 0-12 are explicitly sequenced with validation checkpoints — Step 11 runs the test twice for idempotency, runs the snip command, and includes a bounded fix-retry feedback loop (max 3 attempts) before stopping. | 3 / 3 |
Progressive Disclosure | Overview body cleanly offloads detail to three real one-level-deep reference files (references/page-scanning.md, references/data-adaptation.md, references/extension-handoff.md, all present) with well-signaled 'Read the reference file X for the full...' navigation. | 3 / 3 |
Total | 11 / 12 Passed |