Content
81%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 tight, well-structured recipe: complete input table with defaults, one fully specified generation step with a copy-ready prompt, operational fallbacks, and clear trigger keywords. The only improvements are trimming the duplicated description line and showing the exact CLI invocation syntax for the listed parameters.
Suggestions
Remove the bold line duplicating the frontmatter description directly under the title to save tokens without losing information.
Show the literal `muapi video from-image` command line (with actual flags for image, aspect ratio, duration, and resolution) instead of a parameter:value spec so the step is copy-paste ready.
Consider moving the ~40-line video prompt template into a reference file (e.g. references/prompt-template.md) so SKILL.md stays a lean overview while the prompt is still one level deep.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence — no concept explanations, and the long prompt block is the skill's actual deliverable rather than padding. Minor waste keeps it below anchor 5: the frontmatter description is restated verbatim in bold under the title, and the generic "Notes for the Executing Agent" boilerplate adds little that is specific to this recipe. | 4 / 5 |
Actionability | Concrete and near-executable: a named CLI command with model/quality/audio/resolution specified, a fully written prompt with {{placeholder}} substitution, defaults for every input, a fallback model, and a curl fallback for endpoints without aliases. It falls short of anchor 5 only because the parameters are given as a spec rather than a literal copy-paste command line, and the curl fallback uses a literal '{...}' body. | 4 / 5 |
Workflow Clarity | This is a single-task skill and the single action is unambiguous — submit one generation with fully specified parameters, then present the video — and it even includes an explicit error-recovery path ("If the model rejects realistic human likeness, switch to the Global tier model"). The simple-skill exception applies since generation is not a destructive or batch operation requiring validation. | 5 / 5 |
Progressive Disclosure | The file is well-sectioned (Inputs, Steps, Notes, Trigger Keywords) with no nested or buried references, and no bundle files exist so nothing is hidden. It misses anchor 5 because the body exceeds 50 lines and the ~40-line prompt template could plausibly live in a reference file, leaving the SKILL.md as a leaner overview. | 4 / 5 |
Total | 17 / 20 Passed |