Content
80%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 concise, well-structured overview that assumes Claude's competence and points to package docs and examples for depth. Its main gap is workflow clarity: the feature-selection decision the skill exists for is never given concrete criteria or a sequence.
Suggestions
Add decision criteria or a short decision table in 'When To Use' for choosing between contextMap, contextPolicy, optimization, and recall (e.g., what signals route a task to GEPA vs. memory recall), instead of only listing the four options.
Include one runnable code snippet for at least one of the four core features (e.g., setting a contextPolicy or invoking AxGEPA) and define how `llm` is constructed so 'Core Pattern' is copy-paste ready.
Replace the circular bullet 'Route long-context agent work to the right generated-package feature' with a concrete routing rule or checkpoint that tells the agent when it has chosen correctly.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and fact-dense — bullet-style 'Package Facts' ('Real network support: yes.', 'Runtime profiles: javascript-quickjs, python-pyodide'), a five-line core pattern, and guardrails with no padding or explanation of concepts Claude already knows. Every token earns its place. | 5 / 5 |
Actionability | 'Core Pattern' provides real, runnable-shaped code ('helper = agent("question:string -> answer:string")') and 'Relevant API Surface' lists exact symbols, but `llm` is left undefined and none of the four core decision features (contextMap, contextPolicy, optimization, recall) get a code example — the skill defers to 'examples/' instead. Concrete code with minor gaps, so above 3 but short of copy-paste-ready 5. | 4 / 5 |
Workflow Clarity | 'When To Use' bullets and guardrails give coherent structure, but the skill's central task — choosing between contextMap, contextPolicy, optimization, and recall — has no decision criteria, sequence, or checkpoints; 'Route long-context agent work to the right generated-package feature' restates the purpose without saying how to choose. Not 4 because checkpoints/criteria for the core decision are missing; above 2 because guardrails do constrain behavior. | 3 / 5 |
Progressive Disclosure | At under 50 lines with no bundle files, the body is a well-organized overview with clearly signaled one-level references ('Package API docs: API.md and axir-api.json', 'Runnable examples: examples/') and easy navigation across its short sections, matching the simple-skill exception. | 5 / 5 |
Total | 17 / 20 Passed |