Content
92%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable, well-sequenced analysis workflow with concrete tool calls, thresholds, and a detailed report template. Its only weakness is progressive disclosure: everything is inline with no bundle references, and at over 50 lines it exceeds the simple-skill exception.
Suggestions
Move the ~90-line report output template (lines 105–195) into a references/REPORT_TEMPLATE.md and link to it from Step 5, keeping only the section skeleton inline in SKILL.md.
Extract "Common Patterns" (Churn Risks / Expansion Signals) into a references/PATTERNS.md referenced once near Step 1's classification, reducing body length and improving discoverability.
Add a brief validation note in Step 0 or Step 1 for what to do when Amplitude returns no data or the account property is missing (e.g., ask user to confirm the identifier) to make the data-gap guidance an explicit checkpoint.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and dense—specific tool calls, query parameters, and thresholds with no padding explaining concepts Claude already knows; even the long output template earns its tokens by defining report structure. | 3 / 3 |
Actionability | Concrete, executable guidance throughout: named Amplitude tools, explicit query fields, numeric classification thresholds (DAU/MAU >40%, <20%), and a copy-ready report template. | 3 / 3 |
Workflow Clarity | Steps 0–5 are clearly sequenced with parallelization noted and a health-classification checkpoint that drives conditional branching in Step 3; the task is read-only so the destructive/batch validation cap does not apply. | 3 / 3 |
Progressive Disclosure | Well-organized into clear sections but entirely inline at ~210 lines with no external references; the large output template and Common Patterns are content that could plausibly live in a separate reference file. | 2 / 3 |
Total | 11 / 12 Passed |