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 well-structured, highly actionable skill body with a clear validated workflow and proper progressive-disclosure signaling. The main gaps are minor persona verbosity and the fact that the referenced reference/script bundle files are absent from the reviewed package.
Suggestions
Ship the referenced bundle files (references/prompt-engineering-blog.md, references/mcp-tools.md, references/gemini-models.md, scripts/setup_image_mcp.py, scripts/validate_image_setup.py) so the on-demand references actually resolve.
Trim the 'You are a Creative Director' introduction and a few explanatory asides to lift conciseness toward level 5.
Consider providing one fully-filled example prompt alongside the placeholder templates to push actionability to copy-paste-ready level 5.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly dense and actionable (tables of ratios, MCP tools, error resolutions, executable magick/yaml blocks), but the 'You are a Creative Director' persona intro and a few explanatory asides are minor over-explanation that could be trimmed, landing just above the midpoint rather than fully lean. | 4 / 5 |
Actionability | Provides concrete, executable guidance throughout — real bash commands ('magick input.png -resize 1200x630^ ...'), yaml frontmatter snippets, and named MCP tools with parameters — but the prompt templates rely on intentional placeholders ([shot type], [subject with physical detail]), which are justified flexibility rather than copy-paste-ready, so it sits below level 5. | 4 / 5 |
Workflow Clarity | The generation workflow is a clearly numbered 7-step sequence with an explicit MCP-availability validation checkpoint up front, a safety-filter auto-rephrase feedback loop (max 3 attempts) for error recovery, and a dedicated error-handling table, matching the level-5 anchor with validation steps and feedback loops. | 5 / 5 |
Progressive Disclosure | Structure and signaling are good: details are deferred to one-level-deep references with an explicit 'Load on-demand - do NOT load all at startup' note. However, the referenced bundle paths (references/prompt-engineering-blog.md, references/mcp-tools.md, references/gemini-models.md, scripts/setup_image_mcp.py, scripts/validate_image_setup.py) do not exist in the bundle, so navigation cannot fully resolve, keeping it below level 5. | 4 / 5 |
Total | 17 / 20 Passed |