Content
82%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 lean, well-guarded instructional skill: it enforces reproducibility and anti-fabrication constraints, gives a clear 6-step method with an executable minimal example, and closes with common mistakes. Weaker spots are the sketched (uncoded) income-model guidance and the absence of an explicit simulation sanity-check step.
Suggestions
Add a small executable snippet or concrete parameter template for the Step 4 income models (Markov transition matrix example, lognormal median/sigma defaults marked as 示例假设) so the recommended approach is copy-paste ready.
Include a verification step after simulation, e.g. re-run with 2–3 different seeds and confirm P10/P50/P90 and depletion probability are stable within tolerance before reporting.
Consider moving the income-modeling details and the extended example into a references/ file, keeping SKILL.md as a lean overview with a clearly signaled pointer.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every section carries domain-specific value — the recurrence formula, the 自然到账 vs 提取 cashflow distinction, reproducibility requirements, anti-fabrication guardrails — with no padding or explanation of concepts Claude already knows. The 90-odd lines are all load-bearing. | 5 / 5 |
Actionability | The Python example is fully executable with defaults and the core recurrence and reverse-solve objective ("f(E) = median(A_T(E)) - target", binary/grid search) are concrete. However Step 4's recommended income models (Markov state transitions, lognormal payments) are named without any code or calibration guidance, leaving minor gaps. | 4 / 5 |
Workflow Clarity | Steps 1–6 are clearly sequenced, the Quick Reference acts as a 5-item output checklist, and the reproducibility requirements (N, seed, distribution params) function as validation checkpoints for simulation claims. Minor gap: no explicit sanity-check step for simulation results, such as verifying stability across seeds. | 4 / 5 |
Progressive Disclosure | The skill is self-contained with no bundle files and well-organized sections (Overview, When to Use, Guardrails, Quick Reference, Method, Example, Common Mistakes). Most content is appropriately placed, though the income-model details (Step 4) and possibly the example could live in a reference file to slim the main body. | 4 / 5 |
Total | 17 / 20 Passed |