MCP server for searching and retrieving trail information from AllTrails
Two tools with basic schemas and descriptions, but significant gaps in parameter documentation and output structure. Both tools have descriptions (10-20 chars above minimum) and input schemas with required fields, but lack parameter descriptions, output schemas, and error guidance. No tool annotations (readOnlyHint, etc.). Naming is verb-forward (search_, get_) but descriptions are minimal. No pagination support despite search returning up to 15 results. Error handling returns plain text without recovery guidance.
Get detailed information about a specific trail by its AllTrails slug
Search for trails in a specific national park using AllTrails data
Parameter descriptions missing. 'park' and 'slug' parameters lack descriptions explaining format, constraints, or examples. LLMs cannot infer valid input without explicit guidance.
No output schema documented. Tools return TextContent with markdown-formatted strings, but LLMs need structured field definitions to parse results reliably and chain to downstream tools.
Error responses lack recovery guidance. Errors return plain text ('No trails found', 'Trail not found') without suggesting next steps (e.g., 'Try a different park slug format' or 'Call search_trails to discover valid park slugs').
No tool annotations. Tools lack readOnlyHint, destructiveHint, or idempotentHint metadata. Both are read-only (safe to retry), but this is not declared in the schema.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | 2026-07-28+ | v2 |
search_trails hardcodes result limit to 15 without pagination. Large result sets are truncated silently ('... and N more trails'). No limit parameter, offset, or next_cursor for pagination.