Loading ToolBench
Merriam-Webster MCP — wraps the Merriam-Webster Collegiate Dictionary and Collegiate Thesaurus APIs (dictionaryapi.com). The authoritative American English dictionary and thesaurus. Look up word definitions, part of speech, pronunciation, and etymology, or find synonyms and antonyms.
Two tools with complete schemas and reasonable descriptions, but significant gaps in parameter documentation and error handling guidance. Tool names lack action verbs (define/thesaurus are nouns, not verb_noun pattern). Descriptions are adequate (150-200 chars) but lack WHEN-to-use context and recovery guidance. The _apiKey parameter is exposed as a tool input, violating secret-injection pattern. No output schema documentation. Error responses return raw status codes without actionable recovery hints.
Look up a word in the Merriam-Webster dictionary — the authoritative American English dictionary. Returns word definitions (short definitions), part of speech (noun, verb, adjective, etc.), pronunciation, and etymology (word origin). When the word is not found, returns close spelling suggestions instead. Example: define({ word: "ephemeral" })
Look up a word in the Merriam-Webster thesaurus — the authoritative American English thesaurus. Returns synonyms and antonyms grouped by sense, along with part of speech (noun, verb, adjective, etc.) and short definitions. When the word is not found, returns close spelling suggestions instead. Example: thesaurus({ word: "happy" })
Tool names lack action verbs. 'define' and 'thesaurus' are nouns; should be 'lookup_definition' or 'get_word_definition' and 'lookup_synonyms' or 'get_word_synonyms'. LLMs infer intent from verb_noun pattern.
API key exposed as tool parameter (_apiKey). Violates secret-injection pattern, credentials in tool params leak into logs and traces. Should use server-side environment variable injection only.
Error responses lack recovery guidance. Returns {error: 'api_key_required', message: '...'} and {error: <status>, message: <text>} without telling LLM what to do next. Should include actionable hints like 'Provide your own Merriam-Webster keys via _apiKey parameter' or 'Retry with exponential backoff'.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | 2026-07-28+ | v2 |
Output schema not documented. Code returns structured objects (entries array with id, part_of_speech, pronunciation, definitions, etc.) but no schema is visible in tool definitions. LLMs cannot plan downstream calls without knowing response structure.
Parameter 'word' lacks format/constraint documentation. Description says 'e.g. ephemeral' but does not specify: min/max length, allowed characters, case sensitivity, or whether multi-word phrases are accepted.