Content
77%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 body delivers exceptionally actionable, tool-specific guidance with well-sequenced validation loops and safety gates on destructive operations — its strongest dimensions. It is weakened by redundancy (timing tables and summary/best-practice sections that restate earlier content) and by a progressive-disclosure structure whose referenced guide files are absent from the bundle while detailed reference material stays inlined in SKILL.md.
Suggestions
Merge the 'Quick Reference' and 'Performance Characteristics' tables (they duplicate the same response-time data) and cut or fold the 'Summary' and 'Best Practices' sections, which restate the Common Mistakes table — this alone would remove roughly 100 lines.
Ship the referenced guide files (SEARCH_GUIDE.md, VALIDATION_GUIDE.md, WORKFLOW_GUIDE.md, OPERATIONS_GUIDE.md) in the skill bundle or remove the links — as delivered, all four of the body's primary references point to files that do not exist.
Move the detailed 'Running Workflows' and 'Version History' sections into WORKFLOW_GUIDE.md (keeping only a one-line pointer plus their safety warnings in SKILL.md) so the overview stays lean and the deep detail lives one level down.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with tool-specific facts Claude would not know, but the same facts are restated repeatedly: the 'Most Used Tools' and 'Performance Characteristics' tables duplicate the same response-time data, and the 'Summary' and 'Best Practices' sections largely restate the Common Mistakes table and Tool Selection patterns. This sits between anchor 2 (several padded/duplicated sections) and anchor 4 (minor trimming); the roughly 150 lines of duplication plus scattered version-specific numbers (e.g. 'httpRequest is at 4.4+', 'n8n 2.97.0') outside any deprecation section make anchor 3 the best fit. | 3 / 5 |
Actionability | Calls are given in concrete, copy-paste-ready form with exact parameter shapes: 'search_nodes({query: "slack"})', 'get_node({nodeType: "nodes-base.slack", mode: "docs"})', WRONG/CORRECT pairs for nodeType prefixes and credential nesting, the six required parameters of n8n_explore_node_resources, and exact pinData keying ('{"Webhook": [{"json": {"id": "123"}}]}'). It matches anchor 5: executable patterns covering the common cases, with only cosmetic shorthand (JavaScript-style rather than full JSON envelopes). | 5 / 5 |
Workflow Clarity | Multi-step processes are explicitly sequenced with validation checkpoints and feedback loops: 'validate → fix → validate', create → validate → update → validate again → activate, and per-method tables for running workflows with error-code recovery paths (STALE_CONFIG, EXECUTION_FAILED vs OFFICIAL_MCP_ERROR). Destructive and batch operations carry explicit guards ('use dryRun: true before bulk changes', 'confirm with the user before dropping a populated column', 'ask the user before passing exposeToMcp'), matching the score-5 anchor including its destructive-operation requirements. | 5 / 5 |
Progressive Disclosure | The four guide references (SEARCH_GUIDE.md, VALIDATION_GUIDE.md, WORKFLOW_GUIDE.md, OPERATIONS_GUIDE.md) are one level deep and clearly signaled, but none of these files exist in the bundle, so the disclosure structure is nominal rather than real. In addition, long reference-grade sections (Running Workflows, Version History, Agents — each ending in 'See WORKFLOW_GUIDE.md for runnable examples') are inlined in a ~480-line SKILL.md, fitting anchor 3 ('content that should be separate is inline') rather than anchor 4's 'most content appropriately placed'. | 3 / 5 |
Total | 16 / 20 Passed |