Content
72%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-organized overview skill with excellent progressive disclosure — an index body pointing to three references and three verified scripts — and mostly dense, actionable tables. Its weaknesses are the implicit, scattered workflow (no numbered sequence with validation for the core deliverables) and minor padding (Keywords section, generic chain-of-thought guidance).
Suggestions
Present the core deliverable workflows (financial model, fundraising package, board section) as explicit numbered steps with validation checkpoints — e.g., 1. read company-context.md, 2. ask the Key Questions, 3. build via the reference/scripts, 4. verify against the Red Flags list — instead of leaving the sequence implicit across the Key Questions, Output Artifacts, and Communication sections.
Add a sample invocation with arguments and a short excerpt of expected output for each Quick Start script so the commands are self-contained without opening the script files.
Trim the Keywords section (it duplicates the frontmatter description's triggers) and the generic 'Reasoning Technique: Chain of Thought' paragraph to reduce token cost without losing signal.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly dense, high-signal content — terse tables with specific targets ("Burn Multiple < 1.5x Monthly", "CAC Payback < 18 mo") and no explanations of concepts Claude already knows. Not a 5 because the Keywords section duplicates the frontmatter description's triggers, and the generic "Work through financial logic step by step" chain-of-thought guidance is padding that could be trimmed. | 4 / 5 |
Actionability | Copy-paste-ready Quick Start commands ("python scripts/burn_rate_calculator.py" etc., all verified to exist) and a concrete request→deliverable mapping ("'How much runway do we have?' → Runway model with base/bull/bear scenarios"). Not a 5 because the commands lack arguments or expected-output examples, and directives like "Be conservative in projections — model the downside first" are direction rather than executable guidance. | 4 / 5 |
Workflow Clarity | A rough sequence exists (read company-context.md → ask Key Questions → produce the Output Artifact → pass the Internal Quality Loop → formatted output), and the Quality Loop is a genuine checkpoint, but the sequence is scattered across five sections rather than presented as a workflow, and validation for the main deliverables (modeling, fundraising, board prep) is only implicit — deferring to reference files. This matches anchor 3 ('sequence present but checkpoints missing or implicit') better than anchor 4's 'clear sequence with most checkpoints present'. | 3 / 5 |
Progressive Disclosure | The body is a genuine overview: a Core Responsibilities table mapping each area to its reference file, a Resources section annotating all six bundle files (each verified to exist), one-level-deep references, and no bulk detail inlined. It does not fit anchor 4 because there are no real organization gaps — navigation is easy and every in-bundle path resolves (only the cross-skill '../agent-protocol/SKILL.md' pointer sits outside the bundle). | 5 / 5 |
Total | 16 / 20 Passed |