Server has 5 tools with complete input schemas and descriptions. Naming follows verb_noun convention (search_, add_, get_). However, descriptions are generic (avg 65 chars, below 194-char baseline) and lack context on WHEN to use each tool or what distinguishes similar tools. Parameter descriptions are minimal. No output schema documentation. Error handling is basic (Zod validation only). Missing tool annotations (readOnlyHint, destructiveHint). No guidance on pagination limits or result caps.
Add a new bookmark to Raindrop.io
Get the most recent bookmarks from Raindrop.io
Get all tags from Raindrop.io
Search for bookmarks in Raindrop.io by query or tags
Search for bookmarks in Raindrop.io by tag
Descriptions lack context on tool selection and differentiation. 'Search for bookmarks in Raindrop.io by query or tags' (65 chars) does not explain WHEN to use search_bookmarks vs search_bookmarks_by_tag, or what each returns. Baseline is 194 chars with clear intent.
No output schema documentation. Callers cannot predict response structure (e.g., does searchBookmarks return {items: [], count: number} or {bookmarks: []}?). LLMs must infer structure from JSON responses, risking parsing errors.
Missing tool annotations. add_bookmark is destructive (creates state) but lacks destructiveHint. search_* and get_* tools are read-only but lack readOnlyHint. Agents cannot determine retry safety or side-effect risk.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 49 | 2026-07-28+ | v2 |
Parameter descriptions are minimal or absent. 'page' and 'perPage' lack guidance on valid ranges, defaults, or behavior at boundaries. 'query' in search_bookmarks does not explain if it searches title, tags, or full text. Baseline is 72 chars per param.
No error recovery guidance. Zod validation errors are caught but responses do not suggest next steps (e.g., 'Invalid page number. Must be >= 1'). Agents receive raw error messages with no actionable path forward.