Yelp MCP — wraps the Yelp Fusion API (api.yelp.com/v3). Find local businesses and restaurants, get business details, and read recent Yelp reviews with star ratings.
Three tools with complete schemas and descriptions. Tool names follow verb_noun pattern (search_, get_). Descriptions are adequate (150-200 chars) and include examples. All parameters have types and descriptions. However, output schemas are not documented, LLMs cannot predict response structure. Parameter descriptions lack format constraints (e.g., 'latitude must be -90 to 90'). The _apiKey parameter violates secret-injection pattern by exposing API keys as tool parameters. Error handling returns raw HTTP status codes without recovery guidance. No pagination support despite search returning potentially large result sets.
Get full details for a single Yelp business by its id or alias. Returns rating, review count, price, categories, address, phone, hours, photos, and coordinates. Example: get_business({ id: "garaje-san-francisco" }).
Get recent Yelp reviews and star ratings for a business by its id or alias. Returns review text, rating, author, and timestamp. Example: get_reviews({ id: "garaje-san-francisco", limit: 3 }).
Search for local businesses and restaurants on Yelp by term and location. Returns matching businesses with star ratings, review counts, price, categories, address, and phone. Example: search_businesses({ term: "pizza", location: "San Francisco, CA" }).
API keys exposed as tool parameters (_apiKey). Credentials must use server-side secret injection via environment variables. Agent traces log all parameters, secrets leak into logs and prompt history.
Output schemas not documented. LLMs cannot predict response structure (fields, types, nesting). search_businesses returns {total, businesses: [{id, name, rating, ...}]} but this is invisible to the LLM. Agents must guess what fields exist.
Error handling returns raw HTTP status codes and text without recovery guidance. Example: {error: 401, message: '...'} tells the LLM nothing about whether to retry, ask the user, or give up. Should return 'Invalid API key. Check your Yelp Fusion credentials.'
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | 2026-07-28+ | v2 |
search_businesses lacks pagination. No offset/limit handling for large result sets. Returning all 50 results every time wastes tokens and risks context window exhaustion. Should support limit (1-50, default 20) and offset/cursor.
Parameter descriptions lack format constraints. 'latitude' and 'longitude' have no range hints (-90 to 90). 'limit' has no min/max. 'price' format ('1,2,3') is vague. LLMs cannot validate inputs without explicit constraints.