Model Context Protocol (MCP) server for the Brandfetch API. Enables searching for brands by name and retrieving detailed brand information including logos, colors, fonts, and company details.
The server defines 2 tools with clear verb-noun naming (search_brands, get_brand_info) and complete input schemas with proper typing. However, descriptions lack depth and guidance for LLM selection, parameter descriptions are generic, output schemas are undocumented, and error handling provides minimal recovery guidance. The tools follow basic structure but fall short of production-grade patterns for agent reasoning.
Get detailed brand information by identifier using the Brandfetch Brand API. Args: identifier: Identifier to retrieve brand data. Accepted formats: - Domain: nike.com - Brand ID: id_0dwKPKT - ISIN: US6541061031 - Stock Symbol: NKE fields: Optional list of fields to include in the response. If None, returns all fields. Returns: Detailed brand information, optionally filtered by the 'fields' parameter. The response includes detailed brand assets such as: - Brand name and description - Logos in various formats - Brand colors - Fonts used by the brand - Social media links - Company information
Search for brands by name using the Brandfetch Search API. Args: name: The name of the company you are searching for. client_id: Optional client ID for the API. If not provided, will use the one from environment. Returns: A list of matching brands with their icon, name, domain, claimed status, and brand ID. Example: [ { "icon": "https://example.com/icon.svg", "name": "Example Company", "domain": "example.com", "claimed": true, "brandId": "id_12345" } ]
Output schemas are not documented. Tool descriptions mention what fields are returned (icon, name, domain, claimed, brandId for search_brands; brand name, logos, colors, fonts, social media, company info for get_brand_info) but no formal JSON Schema is provided. LLMs cannot reliably extract or plan around unspecified output structures.
Parameter descriptions are minimal and lack actionable constraints. The 'name' parameter in search_brands has description 'The name of the company you are searching for' (56 chars) with no guidance on format, minimum length, or handling of special characters. The 'identifier' in get_brand_info lists accepted formats but does not explain when to use each or how to resolve ambiguity if multiple match. Descriptions should be 50-200 chars with explicit constraints.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 56 | <=2025-11-25 | v2 |
| 2026-03-09 | D | 54 | - | v1 |
Error handling is generic and non-actionable. Code catches HTTP errors and general exceptions, returning 'Failed to search brands: HTTP <code>' or 'Failed to search brands: <error>'. This tells the LLM nothing about recovery: is the error retryable? Is the identifier invalid? Does the API key need rotation? Errors should classify as retryable/user-fixable/fatal and suggest next steps.
Tool descriptions lack guidance for LLM selection and when to use each tool. search_brands returns a list of brands; get_brand_info retrieves full details for one. The descriptions do not explain: 'Use search_brands first to find a brand by name, then pass the brandId to get_brand_info for full details.' This dependency is implicit, forcing the LLM to infer the workflow.
The 'fields' parameter in get_brand_info accepts any array of strings with no enum constraint or documentation of valid field names. Description states 'Optional list of fields to include in the response. If None, returns all fields.' but does not list what fields are available. An LLM must guess: 'brand_name', 'logo', 'colors', 'fonts'? This invites invalid requests and wasted calls.
No pagination support. search_brands returns 'a list of matching brands' with no mention of limit, offset, or total count. If a query matches 500 brands, the API response could be massive and blow the context window. Tools should accept page/limit and document max result size.
No idempotency guarantees. If search_brands or get_brand_info fails mid-response, an agent may retry and get different results (newly added brands, changed info). Tools that query external APIs should document whether they are idempotent and when to retry safely.