Content
100%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.
An exemplary instruction-only skill body: lean, fully executable, with a well-sequenced per-command workflow, explicit error and authority rules, and appropriate self-contained structure for its size. The only improvement vector is upstream (the description), not the body itself.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every line earns its place: a 2-line context preamble that establishes the requested-vs-executed distinction, a copy-paste CLI resolver, a terse argument-to-command map, and error-handling rules. No padding, no explaining concepts Claude already knows, no verbosity anywhere. | 5 / 5 |
Actionability | Fully executable guidance: the bash block resolves the CLI with concrete fallback paths and a working install hint, and every command maps to an exact invocation (`"$CLI" graph list`, `graph show "<name>"`, `graph new "<request>"`, `graph run "<name>" -y` with `--input "<value>"`). Output expectations per command are specified, covering all cases the description promises. | 5 / 5 |
Workflow Clarity | The run flow is explicitly sequenced with checkpoints: show the graph first, obtain any missing required input without inventing it, then run with -y; a single user run request is the authority. Error handling is defined ("On any non-zero exit, relay the exact refusal and stop; do not retry with a guess") and the interview flow for `new` specifies that the save confirmation is approval. Each sub-command's sequence is unambiguous. | 5 / 5 |
Progressive Disclosure | Under 50 lines with no need for external reference files, and no bundle files exist to link. The two sections ("Resolve the independent Agentlas CLI", "Commands") are well-organized, flat, and each fully delivers its content inline — matching the rubric's exception for short single-purpose skills. | 5 / 5 |
Total | 20 / 20 Passed |