Content
57%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 dense, well-organized strategic reference with concrete commands, metric thresholds, and a quality-loop protocol, weakened by redundancy (duplicated keywords and reference listings), generic CoT padding, soft rather than explicit validation checkpoints, and references to bundle files that are not present.
Suggestions
Remove the 'Keywords' section (it duplicates the description) and consolidate the 'Core Responsibilities' and 'Resources' sections so each reference file is listed once.
Replace the generic 'Chain of Thought...show all math' guidance with skill-specific validation gates, e.g. a numbered workflow for model building that requires downside-first scenarios and explicit validate-before-act checkpoints for cuts/fundraising/board outputs.
Bundle the referenced files (references/financial_planning.md, references/fundraising_playbook.md, references/cash_management.md and the three scripts/*.py) so the signaled navigation actually resolves, or inline the essential mechanics and drop the dangling references.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly dense and token-efficient (tables of metrics, targets, red flags) without explaining concepts Claude already knows, but the 'Keywords' line duplicates the description, 'Core Responsibilities' and 'Resources' re-list the same reference mapping, and the generic 'Chain of Thought...Show all math' section states guidance Claude already applies. | 3 / 5 |
Actionability | Quick Start gives copy-paste executable commands ('python scripts/burn_rate_calculator.py') and the Metrics/Red Flags/Proactive Triggers sections give concrete thresholds and named artifacts, but the core modeling mechanics are offloaded to referenced files rather than shown inline. | 4 / 5 |
Workflow Clarity | A loose sequence exists (Key Questions -> metrics -> conservative reasoning -> Internal Quality Loop) with a verify/tag-confidence feedback loop, but for a skill driving consequential decisions (budget cuts, fundraising, board outputs) the validation checkpoints are soft review steps rather than explicit technical validate-then-proceed gates, triggering the cap. | 3 / 5 |
Progressive Disclosure | The body is well-organized with one-level-deep, clearly-signaled references (a Core Responsibilities 'Reference' column plus a Resources section), but the referenced bundle files (references/*.md, scripts/*.py) do not actually exist on disk, so navigation points to missing destinations. | 3 / 5 |
Total | 13 / 20 Passed |