Content
38%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 command documentation is genuinely actionable, but the skill is buried under large off-topic sections (agent-conduct rules, trading-risk policy) and duplicated, version-conflicting content, with a dangling reference to a nonexistent file. It needs aggressive trimming and restructuring to function as a skill overview.
Suggestions
Remove or move the ANTI-PATTERN REGISTER and the trading/meme-coin risk sections out of SKILL.md entirely — they are agent-conduct policy, not design-engine instructions, and consume most of the token budget.
Merge the "Manuale d'uso approfondito — gofrontai v3.1" with the earlier "Commands" section into one authoritative, de-duplicated command reference, reconciling the v2.0/v3.1 version conflict.
Create the referenced "references/anti-slop-directives-2026.md" file (or remove the pointer) and move the detailed anti-slop workflow and corpus list into it, keeping only a short summary plus the link in SKILL.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Roughly half the body is off-topic padding for a design-engine skill: a ~90-line ANTI-PATTERN REGISTER about delivery quality, financial-data honesty, and meme-coin/trading-risk rules, plus a "Manuale d'uso v3.1" that duplicates the earlier command list. The core design content could fit in a third of the space. | 1 / 5 |
Actionability | Commands are concrete and copy-paste ready ("gofrontai wireframe \"brief\"", "go run ./cmd/design-page \"brief\"", "gofrontai compete --name \"readmago\" --fidelity 8.0") and cover common cases, though no build/setup steps for the binary are given and the duplication plus planned-vs-shipped confusion creates minor ambiguity. | 4 / 5 |
Workflow Clarity | The anti-slop section gives a 5-step pre-render sequence with a re-audit loop, but the document as a whole has no coherent end-to-end workflow: commands appear twice in conflicting forms (v2.0 subcommands vs. v3.1 manual), and there is no validation step for the generated output beyond the anti-slop re-audit. | 3 / 5 |
Progressive Disclosure | The body is a monolithic inline document — the anti-pattern register and v3.1 manual clearly belong in separate reference files — and its one clearly-signaled reference, "references/anti-slop-directives-2026.md", does not exist in the bundle (no references/ directory), nor do the other cited files (ANTI_SLOP_REPOS.md, CLAIMS-VERIFICATION.md, DESIGN_COMPONENT_LIBRARY_PLAN.md). | 2 / 5 |
Total | 10 / 20 Passed |