Content
75%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 strong, highly actionable reference skill with excellent executable examples, SymPy-specific pitfalls (ConditionSet vs EmptySet, `==` vs `Eq`, parse_expr security), and well-signaled one-level-deep references. Its weaknesses are redundancy (duplicated examples and file listings, an explanatory Overview paragraph) and a confusing duplicate core-capabilities file pair with inconsistent naming.
Suggestions
Conciseness: remove duplicated content — the 'Getting Started Examples' section repeats examples already shown (the same quadratic solved in Pattern 1) and 'Reference Files Structure' duplicates the file listing from 'Core Capabilities'; keep one canonical listing with the 'Load when' guidance, and drop the Overview paragraph defining symbolic mathematics plus the generic 'With Matplotlib' plotting example that just plots a lambdified function.
Progressive disclosure: consolidate the two overlapping core-capabilities files (`core_capabilities.md` hub vs `core-capabilities.md` detailed treatment) into a single consistently named file; the interchangeable hyphen/underscore links in SKILL.md invite broken references and confuse navigation.
Workflow clarity: add in-pattern recovery guidance (e.g., in Pattern 1, what to do when a candidate solution fails verification, such as re-checking assumptions or the solution domain) to close the gap between validation checkpoints and full feedback loops.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly SymPy-specific and code-dense, but contains real padding: the Overview paragraph explains what symbolic mathematics is, the same quadratic (x**2 - 5*x + 6) is solved in both Pattern 1 and Example 1, the reference-file listing appears twice ("Core Capabilities" and "Reference Files Structure"), and the Matplotlib integration example is generic plotting Claude already knows. These duplications go beyond the 'minor instances' of the level-4 anchor, so it sits at level 3. | 3 / 5 |
Actionability | Nearly all guidance is executable, copy-paste-ready Python: installation commands, symbol creation with assumptions, solver selection with domain declarations, lambdify pipelines, and verification asserts (e.g., `assert status == 1` and `assert abs(f(solution[0])) < 1e-10` in the SciPy example). Common cases are covered with concrete, correct code. | 5 / 5 |
Workflow Clarity | Pattern 1 (solve then verify solutions) and Pattern 2 (numbered 4-step symbolic-to-numeric pipeline) present clear sequences with validation checkpoints, and the Troubleshooting section maps common errors to fixes. It falls short of level 5 because the patterns themselves lack in-pattern recovery loops (e.g., what to do when verification fails) and there are no checklists for complex processes. | 4 / 5 |
Progressive Disclosure | Structure is good: SKILL.md is an overview, references are one level deep, and "Load when" guidance signals when to open each file. However, two overlapping files (`core_capabilities.md` hub vs `core-capabilities.md` detailed) use inconsistent hyphen/underscore naming that SKILL.md links to interchangeably, and the file structure is enumerated twice in separate sections — organization gaps that keep it below the level-5 anchor. | 4 / 5 |
Total | 16 / 20 Passed |