Content
88%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 strong, highly actionable body: every section carries unique instance-routing knowledge with executable calls, explicit validation before credential writes, and a complete error-recovery playbook. Its weaker spots are mild repetition of the shared-default and credential-safety rules, inline version-gating information, and no use of reference files despite a 230-line body.
Suggestions
Consolidate the shared-default hazard, which is explained three times (intro, 'How targeting behaves', 'Switching'): state the full mechanism once under 'How targeting behaves' and let the other two sections reference it in one line.
Move the error-envelope table and the INSTANCE_AMBIGUOUS deep-dive into a references/ file (e.g. errors.md) summarized by one line plus a pointer in Quick Reference, keeping the recovery playbook's do-this rows inline.
Place the version-availability note (backend 0.1.120+ / n8n-mcp 2.94.0+) in a short 'Older servers' section paired with its `switch` fallback, so time-sensitive gating information is segregated from the durable guidance.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with n8n-mcp-specific knowledge Claude cannot know (routing semantics, fail-close behavior, error envelopes) and explains nothing generic. Minor trimming opportunities remain: the shared-default hazard is restated in three sections (intro, 'How targeting behaves', 'Switching') and credential-safety rules in four, and the version-availability note ('backend 0.1.120+ / n8n-mcp 2.94.0+') is time-sensitive info placed inline rather than in a compat/deprecated section. Efficient with minor over-explanation → 4, not 5; not 3 since each repetition targets a distinct failure mode. | 4 / 5 |
Actionability | Fully executable throughout: exact call shapes (`n8n_instances({mode:"list"})`, `n8n_manage_credentials({action:"create", …, instance:"client-a"})`), a numbered core workflow, a step-by-step copy-between-instances recipe with read/create/verify calls, concrete JSON error payloads, and a symptom→action recovery table. Copy-paste-ready guidance covering the common multi-instance cases matches the 5 anchor. | 5 / 5 |
Workflow Clarity | Clear sequencing (Golden rules → numbered Core workflow → Recovery playbook) with explicit validation checkpoints at every high-stakes point: 'Check the name against `list` first' before credential writes, 'confirm current.name == "prod"' after a switch, 're-confirm current, then write', and 'verify on B' after a copy. Feedback loops exist for every error code, matching the 5 anchor. The destructive-operation cap does not apply because verification before credential writes is explicitly mandated. | 5 / 5 |
Progressive Disclosure | A single self-contained file with well-labeled sections, a Quick Reference index, and named pointers to sibling skills, so navigation is easy. However, everything is inline — no bundle files or one-level-deep references exist, and at ~230 lines the error-envelope table and the INSTANCE_AMBIGUOUS deep-dive are plausible candidates for a separate reference file. Good structure with minor organization gaps → 4; not 5 because this is not an under-50-line skill where well-organized sections alone suffice. | 4 / 5 |
Total | 18 / 20 Passed |