Content
35%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.
The body reads as an agent-persona manifesto rather than a usable skill: it exhaustively enumerates capability areas, traits, and example prompts while providing no actual analytical content (formulas, benchmarks, templates, or procedures). The description frontmatter does the triggering work; the body adds length without adding knowledge Claude does not already have.
Suggestions
Replace the capability and knowledge-base bullet lists with actual instructional content: concrete TAM/SAM/SOM formulas, a worked unit-economics example (CAC payback calculation), and stage-specific benchmark tables (e.g., burn multiple targets by round).
Cut the padded sections — Behavioral Traits, Example Interactions, When to Use This Agent, and Quality Standards largely restate the description's trigger terms; the body could shrink to a fraction of its length with no loss of guidance.
Move detailed framework material (industry-specific model templates, compensation benchmarks) into reference files under references/ and link them explicitly by path, keeping SKILL.md as a concise overview with clearly signaled one-level-deep references.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~300-line body is dominated by padded sections: "Behavioral Traits" (10 adjective bullets like "Data-driven: Always grounds recommendations in data"), "Knowledge Base" entries that only name topics ("Common pitfalls and how to avoid them" without describing any pitfall), and 18 "Example Interactions" that restate the description's trigger terms. This matches anchor 2 (noticeably verbose, several padded sections) rather than anchor 3, since the padding is the majority of the body rather than isolated instances. | 2 / 5 |
Actionability | The body describes rather than instructs: "TAM/SAM/SOM calculations using bottom-up and top-down methodologies", "Unit economics analysis (CAC, LTV, payback period)", and "Compensation benchmarks (US-focused, 2024)" are all stated without a single formula, benchmark number, model template, or worked example. The "Response Approach" and "Output Format" sections offer high-level process hints, matching anchor 2 (minimal concrete guidance, missing specific steps to execute) rather than anchor 1 only because those two sections give some structural direction. | 2 / 5 |
Workflow Clarity | The "Response Approach" provides a clear 10-step sequence including "Validate findings - Cross-check with benchmarks and alternatives" and "Acknowledge limitations", so a sequence exists. But checkpoints are named without mechanism (no how to cross-check, no error-recovery loop), matching anchor 3 (sequence present but checkpoints implicit) rather than anchor 4's explicit verification commands. | 3 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent), and the body's pointers to detail ("Leverages skills: market-sizing-analysis, startup-financial-modeling...") are bare names with no paths or links. Sections are well-headered, but ~300 lines of capability lists that belong in separate reference files are inlined, matching anchor 3 (some structure, content that should be separate is inline, references not clearly signaled). | 3 / 5 |
Total | 10 / 20 Passed |