Generate LLMs.txt files for websites following the llmstxt.org specification
Single tool with partial schema definition. Tool name follows verb_noun pattern (generate_llms_txt). Input schema present with proper type and description for the 'url' parameter. Output schema declared but lacks granularity on return structure. Description is adequate (106 chars) but generic, does not explain WHEN to use, what happens if generation fails, or prerequisites. Parameter description for 'url' is detailed (102 chars) and helpful, including format variants. However, output schema fields are minimally described (e.g., 'files_content' just says 'Content of generated files' without structure), and no error handling guidance provided in description. Risk category (WRITE) is correctly identified. Missing: per-parameter constraints, error recovery guidance, idempotency statement, and rate-limit warnings for the web-crawling operation.
Generate LLMs.txt files for a website following the llmstxt.org specification
Output schema fields lack detailed descriptions. 'files_content' is documented as 'Content of generated files' without specifying structure (object of filename→content pairs?). 'files_generated' and 'output_directory' are similarly terse.
Tool description does not explain error conditions, recovery paths, or prerequisites (e.g., 'Generation may fail for URLs behind authentication or paywalls. Try public documentation sites first.').
No mention of idempotency or retry safety. Tool modifies filesystem and fetches remote content, agents need to know: is it safe to retry? Will it overwrite existing files?
Error handling in code returns a flat error dict with 'success: false' but does not validate early or provide actionable guidance. E.g., empty URL returns a generic error; invalid URLs (unreachable hosts, timeouts) are not shown to have specific error messages.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 52 | 2025-06-18+ | v2 |
| 2026-03-09 | D | 55 | - | v1 |
Tool description does not mention rate limits, crawl politeness, or expected execution time. Web crawling can be slow or fail silently on large sites, agents should expect delays and potential timeouts.
Parameter 'url' accepts multiple formats but no enum or strict pattern validation shown in schema. Schema allows free-form string; code tries to normalize it, but LLMs may pass invalid formats anyway.