Content
78%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 tight, well-structured two-phase workflow with concrete commands, model IDs, and ready-to-use prompts, plus an explicit user-approval checkpoint. Main weaknesses are the never-shown merge command, the typo-like duplicated alternative in Phase B step 2, and the absence of explicit output validation between generation and final delivery.
Suggestions
Replace the Phase B step 2 duplication ('muapi video from-image or muapi video from-image') with the single intended command, and give the exact ffmpeg concat command (or a script reference) so the merge step is executable rather than a parenthetical note.
Add an explicit validation checkpoint between generation and presentation, e.g. 'Verify each video completed via muapi predict result before concatenating, and re-submit failed generations' — this also strengthens the feedback loop for the batch-like two-shot workflow.
Show one complete CLI invocation (with argument syntax) for muapi image generate and muapi video from-image so the commands are copy-paste ready rather than implied.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean — an inputs table, phased steps with literal prompt text, and terse agent notes; it does not explain concepts Claude already knows. Minor trimmable padding exists (the 'Trigger Keywords' section largely duplicates the description's job, and the final Notes repeat fallback mechanics). Efficient with minor over-explanation — anchor 4, just short of the every-token-earns-its-place anchor 5. | 4 / 5 |
Actionability | Concrete guidance throughout: specific CLI commands with model IDs ('muapi image generate' model='nano-banana-2-edit', 'grok-imagine-image-to-video'), full ready-to-use prompts, and a curl fallback for unaliased endpoints. Gaps: full CLI invocation syntax is never shown, the ffmpeg merge is only a parenthetical 'You can use... ffmpeg concat' with no command, and Phase B step 2 reads 'muapi video from-image or muapi video from-image' — a duplicated non-alternative. Mostly executable with minor gaps — anchor 4, not 5. | 4 / 5 |
Workflow Clarity | Clear sequence: Phase A image generation, an explicit user-approval checkpoint ('Present the luxury image to the user for approval'), Phase B's two timestamped sequential shots, then merging and final presentation. Not 5 because output validation is implicit — there is no step to confirm each generation succeeded or verify the concatenated result before presenting, and the merge step is framed as optional. | 4 / 5 |
Progressive Disclosure | No bundle files exist (no references/, scripts/, or assets/ directories) and the body is self-contained with well-organized sections (Inputs, Steps with Phase A/B, Trigger Keywords, Notes) and no nested references. Per the simple-skill guidance, a short single-purpose skill with no need for external references scores 5 with just well-organized sections, which this has. | 5 / 5 |
Total | 17 / 20 Passed |