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.
An operationally excellent body: concrete tool-call syntax, exact data shapes, explicit validation feedback loops, and hard-won false-positive knowledge Claude could not reconstruct on its own. Its weaknesses are redundancy — timing stats and the iterative-loop message repeated several times, plus a Summary section that recaps the body — and references to three detail files that are not present in the bundle being evaluated.
Suggestions
State the "validation is iterative (2-3 cycles)" fact once and delete its repetitions in the telemetry pattern, the JS example, and the Summary; the timing statistics (23s/58s) add little and can go entirely.
Trim or remove the closing Summary section — six of its key points restate earlier sections verbatim.
Move the four known false-positive issues (full message text, confirmation conditions, issue links) into FALSE_POSITIVES.md and keep only a one-line pointer plus the strictest caveat (don't run default n8n_autofix_workflow on the error-output-config false positive), and ship the three referenced .md files with the skill so the links resolve.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly dense with genuinely non-obvious operational knowledge, but the timing statistics ("23s thinking, 58s fixing", "2-3 iterations") are repeated in at least four places (Validation Philosophy, the telemetry pattern, the JS example, and the Summary), the "validation is iterative" message is restated three times, and the closing Summary section largely re-states earlier sections. Version-pinned details (n8n-mcp 2.63.0/2.85.0, issue #1111's full diagnostic text) are woven through the body rather than isolated, which the guidelines penalize. This matches anchor 3 ("mostly efficient but includes some unnecessary explanation or could be tightened") better than anchor 4's "minor instances". | 3 / 5 |
Actionability | Every section gives copy-paste-ready tool calls with exact parameters: validate_node({...}) with profile settings, n8n_autofix_workflow with applyFixes/confidenceThreshold/fixTypes, a cleanStaleConnections operation, and the pinned-data run path including the exact item shape ("[{\"json\": {...}}]") with the flat-object mistake called out. Error examples arrive with their fix text, and false positives come with precise confirmation conditions and remediation. This is a full anchor-5 match — executable guidance covering the common cases. | 5 / 5 |
Workflow Clarity | The core loop is explicit and re-validated at every stage: validate → read errors → fix → validate again, with recovery strategies sequenced for each failure mode and a dedicated "Running the workflow after it validates" section that walks trigger-detection, pinned-data, consent, and reading EXECUTION_FAILED results. The feedback-loop pattern the rubric rewards is the skill's central organizing principle, and a checklist reference exists for review workflows — a clear anchor-5 match. | 5 / 5 |
Progressive Disclosure | References to ERROR_CATALOG.md, FALSE_POSITIVES.md, and REVIEW_CHECKLIST.md are bolded, one level deep, and consolidated in a "Detailed Guides" section, with the body correctly holding summaries and deferring full per-case detail — matching anchor 4 ("good structure; references mostly clear; minor organization gaps"). It is not anchor 5 because the bundle contains none of the three referenced files (verified: no references/ directory or sibling .md files exist), and some catalog-grade detail (the four known-false-positive issues with full message text and links) is inlined in the body even though FALSE_POSITIVES.md is its designated home. | 4 / 5 |
Total | 17 / 20 Passed |