MCP server for llm.energy - Extract documentation from any site with llms.txt support
The server provides 8 well-named tools with consistent verb-noun naming (extract_, fetch_, get_, list_, verify_, discover_). All tools have descriptions and basic input schemas. However, there are significant gaps in schema completeness and parameter descriptions. No output schemas are documented, LLMs cannot understand what each tool returns. Parameter descriptions are minimal or absent in several tools. Error handling guidance is not visible in the source. Security considerations (e.g., URL validation, rate limiting) are not evident. The implementation is functional but lacks the rigor expected for production agent use.
Discover possible llms.txt locations for a given domain. Checks root domain, docs subdomain, and common paths.
Extract and parse documentation from a website that has llms.txt or llms-full.txt. Returns structured documents ready for AI consumption including individual sections, a full consolidated document, and an agent guide.
Fetch the raw llms.txt or llms-full.txt content from a website without parsing. Useful for quick access to documentation.
Get the AI agent guide (AGENT-GUIDE.md) from a previously extracted URL. This contains instructions on how to use the documentation.
Get a specific document section from previously extracted documentation by filename.
Get the complete consolidated documentation (llms-full.md) from a previously extracted URL.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract specific fields from responses. Pattern requires 'Document the output schema'.
Parameter descriptions are minimal or missing context. 'url' is described only as 'The URL of the documentation website' with no guidance on format (http/https), validation rules, or what happens if invalid. LLM cannot distinguish between a domain and a full path.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-21 | D | 57 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 26 | - | v1 |
List all document sections from a previously extracted URL.
Check if a website has a valid llms.txt or llms-full.txt file. Returns availability status, file location, and basic metadata.
No error handling guidance visible in tool descriptions. Pattern requires 'Error responses must tell the LLM what to do next'. E.g., extract_documentation does not document what happens if the URL has no llms.txt file or if the domain is unreachable.
No pagination or result limiting documented. Tools like list_extracted_documents could return hundreds of items, blowing the context window. Pattern requires capping results at 20-50 items and offering pagination.
Tools for retrieving cached documents (get_document_section, get_full_documentation, get_agent_guide) do not clearly document the relationship to extract_documentation or what happens if the URL was never extracted. Undocumented dependency between tools.
No idempotency guarantee documented. Pattern requires agents to know whether repeated calls are safe. extract_documentation called twice with the same URL, should it return fresh results or cached results? This affects agent retry logic.
No input validation constraints visible. URL parameter has no regex, length limit, or format guidance. LLMs could pass invalid URLs like 'just-a-domain' or 'ftp://example.com' without explicit validation rules stated in descriptions.