Content
85%Weight 40%Scale 1-3Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill body is highly actionable with executable code, exact tool signatures, and a rigorous verification workflow with feedback loops, and it makes good use of one-level-deep references. Its main weakness is length and repetition across sections, which hurts token efficiency without adding new information.
Suggestions
De-duplicate rules that appear in multiple sections (placeholder guidance, IF-branch wiring warnings, 'do not load planning/create-tasks') by stating each once and cross-referencing, to tighten conciseness.
Consider moving the large code-heavy SDK Patterns Reference and Expression Reference into a references file so the main SKILL.md stays a lean overview.
Collapse the repeated escalation/placeholder restatements in Escalation, Placeholders, and Mandatory Process into a single authoritative section.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is densely packed with n8n-specific gotchas Claude would not know (SDK builder subset limits, mock output shapes, n8n credits rules), avoiding generic concept explanation, but at ~800 lines it repeats several rules across sections (placeholder rules, IF-wiring warnings, and 'do not load planning' each restated two to three times), so it could be tightened. | 2 / 3 |
Actionability | It provides executable TypeScript SDK code (workflow, IF/Switch/Merge wiring, import shape), exact tool-call signatures such as `nodes(action="type-definition")` and `workflows(action="get-as-code", workflowId)`, and concrete gotchas (Gmail removeLabels, resource-locator objects, mock output shapes) that are copy-paste ready. | 3 / 3 |
Workflow Clarity | The 11-step Mandatory Process and the Verification Contract give a clear sequence with explicit validation checkpoints and feedback loops (verify -> shouldEdit: true -> one batched repair -> rebuild -> retry -> stop if the failure repeats), including stop conditions for repair budgets. | 3 / 3 |
Progressive Disclosure | Two real one-level-deep references are clearly signaled and verified to exist ("load this skill's references/error-workflows.md" and "references/compositional-workflows.md"), and the inline content is organized into well-headed sections; the genuinely composable topics are appropriately split out. | 3 / 3 |
Total | 11 / 12 Passed |