Content
86%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 tight, well-organized instruction-only skill body: it adds only project-specific conventions, defines each agent behavior as a concrete condition→action rule, and includes a real contradiction-handling checkpoint. Remaining gaps are minor — a concrete stub example (path + frontmatter) and an explicit ordering linking stubbing, linking, and consistency checking would make the guidance fully copy-paste executable.
Suggestions
Add one concrete stub example (a file path plus the frontmatter fields to fill) so the auto-stub behavior is copy-paste executable rather than descriptive.
Make 'Thread the graph' actionable by naming which frontmatter/link fields to populate when linking characters ↔ factions ↔ settings ↔ lore, or fold it into the stub recipe.
State the order of operations when a new mention is found (stub → backlink → link to existing entities → check consistency) so the multi-behavior flow has an explicit sequence.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~24-line body is lean and every statement is project-specific knowledge Claude cannot infer (folder roles, auto-stub policy, the TTRPG exclusion); nothing explains concepts Claude already knows and no section is padded — matching the 'every token earns its place' anchor. It is not a lower score because there is no over-explanation anywhere. | 5 / 5 |
Actionability | Guidance is mostly executable: per-folder template names are enumerated ("Ships `faction`, `political-faction`, and `religion` templates"), stub behavior has a concrete trigger and output ("stub a file in the right folder with backlinks"), and creation uses a concrete call "write({ document: { path, template: \"<name>\" } })". It falls short of level 5 because "Thread the graph" is direction rather than instruction and no concrete path or frontmatter example for a stub is given. | 4 / 5 |
Workflow Clarity | Each behavior is an unambiguous condition→action pair, and contradiction-flagging is an explicit checkpoint with error-handling direction ("surface the conflict — not silently 'fixing'"), which fits 'clear sequence with most checkpoints present'. It is not 5 because the skill has multiple behaviors with no overall ordering, and the Templates section ends without tying template creation into the stubbing flow. | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines with no external reference files needed (no references/, scripts/, or assets/ exist in the bundle) and is cleanly organized into Folders / Agent behaviors / Templates sections, which the rubric's simple-skill guidance scores at 5 for well-organized sections. | 5 / 5 |
Total | 18 / 20 Passed |