Content
78%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 well-built strategy skill body: a coherent 5-layer framework of specific directives, a clear sequenced workflow with failure patterns and a defined output format, and clean progressive disclosure into two real, substantive reference files. The main weaknesses are minor: no inline examples of the most common implementation cases and no explicit validation checkpoint within the workflow's implementation phase.
Suggestions
Add one brief inline example per high-frequency pattern (e.g., a question-header with its direct-answer opening, or a minimal FAQPage JSON-LD snippet) so common cases are executable without opening the reference files.
Insert an explicit validation checkpoint in the workflow between implementation and re-testing — e.g., after schema and llms.txt changes, validate in the Schema.org Validator and Rich Results Test before querying AI products again.
Trim rationale asides ("This skill encodes principles... The principles age slower than the products", "It helps, but it is one of many signals") to reduce token cost without losing any directive content.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Bullets are tight and directive ("Open major sections with a definitive 1 to 3 sentence answer", "Publication date AND last-updated date, both visible AND in schema") with no teaching of basics Claude already knows. A few rationale asides ("This skill encodes principles... The principles age slower than the products", "It helps, but it is one of many signals") could be trimmed. Not 5 because of those minor padded sentences; not 3 because there is no unnecessary concept explanation. | 4 / 5 |
Actionability | Concrete, executable directives throughout: "A markdown file at /llms.txt", named crawlers ("GPTBot, ClaudeBot, Google-Extended, PerplexityBot"), named validators ("Schema.org Validator AND Rich Results Test"), named schema types (FAQPage, HowTo, BreadcrumbList), and a concrete output file ("aeo-geo-strategy.md"). Not 5 because there are no inline copy-paste examples of the common cases (question-headers, a schema snippet) — those live in reference files; not 3 because guidance is specific and executable, not high-level hints. | 4 / 5 |
Workflow Clarity | Six clearly sequenced steps (audit → priority queries → test current visibility → identify gaps → layer-by-layer plan → implement and re-test) with checkpoints ("Score each" layer, "Note which sources they cite", quarterly re-test as a feedback loop). Not 5 because there is no explicit validation checkpoint inside implementation (e.g., validate schema/llms.txt before re-testing AI visibility) and no error-recovery loop; not 3 because sequence and checkpoints are present, and the workflow is not a destructive/batch operation requiring the cap. | 4 / 5 |
Progressive Disclosure | Body is a well-sectioned overview (~156 lines) with two one-level-deep references, both verified to exist on disk and substantive (llms-txt-guide.md at 227 lines, extraction-friendly-patterns.md at 266 lines), clearly signaled both inline ("See references/llms-txt-guide.md") and in a dedicated "Reference files" section with per-file descriptions. Matches the top anchor; not 4 because structure, signaling, and file placement all leave nothing to fix. | 5 / 5 |
Total | 17 / 20 Passed |