Content
63%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-sequenced, highly concrete modeling workflow with specific benchmarks, formulas, and output templates. The main flaws are generic boilerplate sections, a broken reference to a nonexistent playbook file, and inlined report-template content that should live in a separate reference.
Suggestions
Delete the generic filler sections ('Use this skill when', 'Do not use this skill when', the first '## Instructions') that restate the obvious, and merge 'What This Command Does' into the intro to cut roughly 20 lines of padding.
Fix or remove the broken 'resources/implementation-playbook.md' reference — the file does not exist in the bundle — and move the Step 9 report table templates into it so SKILL.md stays a lean overview.
Convert Section 10 validation from a report deliverable into an explicit build-time feedback loop (compute, sanity-check against benchmarks, fix assumptions, re-verify), and fix the corrupted template cell '$Xم'.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The core Steps 1-10 are dense with domain benchmarks and formulas that earn their tokens, but generic filler pads the body: 'Working on financial projections tasks or workflows', 'You need a different domain or tool outside this scope', two separate '## Instructions' sections, and '## What This Command Does' duplicating the intro. It is not 2 because most content is non-redundant domain guidance, and not 4 because several sections are pure boilerplate that could be deleted. | 3 / 5 |
Actionability | Guidance is highly concrete: the cohort MRR formula, cost targets ('SaaS: 75-85%'), scenario deltas ('-30% vs. base'), metric targets ('LTV:CAC ratio (target > 3.0)'), and full report table templates. It falls short of 5 because the pointer to 'resources/implementation-playbook.md' targets a file that does not exist and a stray character corrupts one template cell ('$Xم'). | 4 / 5 |
Workflow Clarity | A clear 10-step sequence runs from gated input gathering (Step 1's explicit checklist) to saving the model (Step 10), with validation named in report Section 10 ('Sanity checks performed, benchmark comparisons') and in the Do list. It is not 5 because validation is an output deliverable rather than an explicit feedback loop (validate, fix, re-check) during the build. | 4 / 5 |
Progressive Disclosure | Section structure is consistent and easy to navigate, but no bundle files exist at all while the body references 'resources/implementation-playbook.md' (nonexistent), and roughly 60 lines of report table templates are inlined that belong in a reference file. It is not 2 because references are clearly signaled rather than buried, and not 4 because the single reference is broken and heavy template content is inlined in SKILL.md. | 3 / 5 |
Total | 14 / 20 Passed |