Content
50%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a thorough, well-referenced GEO audit guide with concrete output and threshold guidance, but it is verbose with inline time-sensitive data, lacks executable how-to-check mechanics and a sequenced validation-backed workflow, and keeps heavy detail inline rather than fully offloaded to references.
Suggestions
Move time-sensitive dates and version numbers into a dedicated 'Last updated / deprecated' section or a reference file, and trim inline statistic tables to the most decision-relevant figures, pushing full evidence tables into references.
Add concrete how-to-check steps (e.g., a snippet to fetch and parse robots.txt, a command to test server-side rendering, how to detect /llms.txt) instead of only 'Check for' lists.
Present the audit as an explicit numbered sequence (fetch URL → robots.txt → llms.txt → citability → structure → brand signals → write GEO-ANALYSIS.md) with a validation checkpoint before generating the report.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is information-rich with domain data Claude lacks, but it is verbose and scatters many time-sensitive dates/versions inline ('May 2026', '2026-06-29', '2026-05-19', '2025-11-20', v2.2.4) outside any deprecated/old-patterns section, which the rubric penalizes. | 2 / 3 |
Actionability | It gives a concrete output spec (10-section GEO-ANALYSIS.md), exact thresholds (134-167 word passages), an llms.txt template, and MCP tool names, but mostly as 'Check for...' lists lacking the how-to-perform mechanics for each check. | 2 / 3 |
Workflow Clarity | Content is organized by topic rather than as a sequenced audit workflow; the Output section is a component checklist with no explicit step sequence or validation checkpoint before report generation. | 2 / 3 |
Progressive Disclosure | Two real one-level-deep references (google-ai-optimization-guide.md, llmstxt-evidence.md) are clearly signaled, but the ~315-line body keeps large statistic and crawler tables inline that could be split into references, so content is not appropriately separated. | 2 / 3 |
Total | 8 / 12 Passed |