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 dense, highly actionable skill body: executable CLI commands with flags and env overrides, an explicit validate-and-retry loop capped at 3 iterations with recovery and exit criteria, and clean one-level-deep reference routing to resources files. Its main weakness is redundancy — duplicated resource lists, outputs stated twice, and Canva/MCP configuration detail inlined that belongs in canva-integration.md.
Suggestions
Merge the 'Dependencies' resource list and the 'References' section into a single deduplicated list (each resource stated once with its purpose) to cut ~10 lines and remove navigation ambiguity.
Keep either the 'Expected outputs' bullets or the YAML outputs block, not both — they describe the same three artifacts twice.
Move the MCP config-file path enumeration (project: .agents/mcp.json, .gemini/settings.json, global: ~/.gemini/antigravity-cli/mcp_config.json, etc.) into resources/canva-integration.md and reference it from the resource-scope table, keeping the font-CDN allowlist detail with fixed-stage.md or the CLI pointer.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body never explains concepts Claude already knows and is information-dense throughout, but it includes avoidable duplication: the resources list appears nearly verbatim in both 'Dependencies' and 'References', the 'Expected outputs' bullets are restated as a YAML outputs block, and inline minutiae like the full MCP config-file path enumeration ("~/.gemini/antigravity-cli/mcp_config.json (agy global)") belong in canva-integration.md. This fits 'mostly efficient but... could be tightened' — not a 4 because the duplication is more than minor trimming, not a 2 because there is no padded explanation or filler prose. | 3 / 5 |
Actionability | The 'Canonical command path' gives copy-paste-ready CLI invocations with real flags (e.g. "oma slide validate --workspace \"$DECK_DIR\" --output json", "oma slide export png --workspace \"$DECK_DIR\" [--resolution 720p|1080p|1440p|2160p|4k]") plus env-var overrides, exact artifact paths, doctor preconditions, and guardrails phrased as concrete rules ("Wrap all CSS animations in @media (prefers-reduced-motion: no-preference)"). This matches 'Fully executable; copy-paste ready code or commands' — not a 4 because there are no missing key details for the common cases. | 5 / 5 |
Workflow Clarity | The Structural Flow gives a clear Entry sequence, explicit Transitions, an enumerated 'Failure and recovery' section, and defined Exit criteria, with an explicit validation feedback loop: "Validate (after writing slides)" → "max 3 auto-fix iterations, then surfaces diff to user" plus a referenced pre-delivery checklist. This matches 'Clear sequence with explicit validation steps; feedback loops for error recovery; checklists for complex processes' — not a 4 because validation checkpoints are explicit at every stage rather than having minor gaps. | 5 / 5 |
Progressive Disclosure | The body is an overview that defers detail to one-level-deep, well-signaled references: a References section maps each resources/*.md file to its purpose ("Stage rules + embed instructions: resources/fixed-stage.md", "Pre-delivery gate: resources/checklist.md"), and no reference chains deeper than one level. Not a 5 because the reference list is duplicated from Dependencies (weaker navigation), and some content that belongs in the referenced files — the MCP config path enumeration and the 16-item guardrails list — is inlined in SKILL.md; not a 3 because references are clearly signaled and the split is otherwise appropriate. | 4 / 5 |
Total | 17 / 20 Passed |