Content
57%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 well organized with concrete, runnable commands and useful agent-facing guidance (JSON output, return codes, output verification). However, it repeats example commands across three sections, never demonstrates the core shape/connector syntax, omits an end-to-end workflow with validation, and inlines what is essentially full API reference material.
Suggestions
Add one end-to-end example (project new, shape add, connect add, export render) with explicit validation steps — check the JSON return code and verify the exported file exists — instead of repeating basic commands in three sections.
Show actual syntax for 'shape add' and 'connect add' (arguments, shape types, IDs) since these are the skill's core operations but currently have only one-line descriptions.
Move the full command tables into a references file (e.g. COMMANDS.md) and keep a quick-start in SKILL.md, replacing the vague 'See README.md in the package' pointers with concrete bundle paths.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The command tables are lean, but 'Basic Commands', 'Examples', and 'Output Formats' repeat the same invocations (project new, REPL, --json project info) three times, leaving clear tightening opportunities. | 3 / 5 |
Actionability | Commands are concrete and executable (e.g. 'cli-anything-drawio --project myproject.json export render output.pdf --overwrite'), but the core shape and connector operations get only one-line table descriptions with no syntax example showing arguments or shape IDs. | 4 / 5 |
Workflow Clarity | Command groups are listed but no end-to-end sequence (create project, add shapes, connect, export) is shown, and destructive/batch operations (remove, --overwrite) lack validation checkpoints — only the 'For AI Agents' tips ('Check return codes', 'Verify outputs exist') gesture at validation. | 3 / 5 |
Progressive Disclosure | Sections and tables are well organized, but ~150 lines of inline command reference could live in a separate file, and the 'More Information' pointers ('See README.md in the package', 'HARNESS.md in the cli-anything-plugin') reference external, non-linkable paths rather than bundle files. | 3 / 5 |
Total | 13 / 20 Passed |