Content
86%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 is a lean, fully executable reference for the keyword/language/embedding surface with genuine tool-specific knowledge (feature gating, score normalization, config-only vs CLI flags). Its one real weakness is bundle structure: every promised deep-dive file is missing from this bundle, so the progressive-disclosure pointers dead-end.
Suggestions
Ship references/advanced-features.md and references/cli-reference.md in this bundle (or drop the references and inline the minimal needed detail), since neither file exists in the current bundle.
Replace the cross-bundle pointer to the sibling skill's references/python-api.md and references/configuration.md with a concrete path (e.g., ../xberg/references/python-api.md) so the location is unambiguous.
Add a one-line check for the feature-gated silent-ignore case (e.g., how to confirm the CLI was built with keywords-yake/keywords-rake) to close the workflow validation gap.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with tool-specific facts — real flags, defaults, config keys, feature-gating behavior, and pitfalls — with no filler or explanation of concepts Claude already knows. Algorithm descriptions are one line each and serve selection, not education. Every section earns its tokens. | 5 / 5 |
Actionability | All guidance is copy-paste ready: complete CLI invocations with jq pipelines ("xberg extract paper.pdf --format json --config-json '{\"keywords\"...}'"), a full TOML config block, executable Python with imports, a provider/flag table, and a pitfalls section that gives exact workarounds ("--config-json '{\"keywords\"...}}'" for the missing flag, "xberg cache warm --all-embeddings" for cold models). | 5 / 5 |
Workflow Clarity | Each feature workflow (enable keywords, detect language, embed text, warm cache, programmatic access) is unambiguous with exact commands, and cross-feature piping is shown ("xberg extract report.pdf | xberg embed --preset quality"). Checkpoints exist for the risky cases — the silent feature-gate behavior and cold-model pitfall are explicitly diagnosed — but there is no hint of how to verify the CLI was built with the keywords features, leaving a minor validation gap. | 4 / 5 |
Progressive Disclosure | Section structure and inline reference signaling are good (each See-reference is one level deep and contextually placed), but the bundle contains no references/ directory at all — "references/advanced-features.md" and "references/cli-reference.md" are dead paths in this bundle, and the other two references point to a sibling skill. Four of four referenced paths cannot be resolved locally, so navigation breaks exactly where deeper detail is promised. | 3 / 5 |
Total | 17 / 20 Passed |