MCP server for searching and retrieving Font Awesome icons with detailed information and SVG markup
The server provides 3 well-named tools that are verb-based (get_, search_) and follow action-oriented conventions. All tools have Zod schemas with type definitions and reasonable parameter constraints. However, tool descriptions lack the specificity and recovery guidance needed for robust LLM operation. Parameter descriptions are generic and do not explain WHY the LLM should use each parameter or when dependencies between parameters apply. Output schemas are not documented, the tools return text-formatted results rather than structured objects, forcing LLMs to parse unstructured Markdown instead of extracting typed fields. Error handling returns raw API errors without recovery guidance. The server is functional for simple queries but lacks the refinement patterns (structured output, error classification, chaining IDs) seen in production-grade tools.
Get detailed information and SVG markup for a specific Font Awesome icon.
Get all available Font Awesome icon styles and families.
Search for Font Awesome icons by query string. Returns matching icons with their names, unicode values, and available styles. Optionally filter by style.
Output schemas not documented. All tools return text-formatted Markdown responses instead of structured JSON objects with typed fields. LLMs must parse unstructured text, wasting tokens and introducing parsing errors.
Error handling does not guide recovery. The search_icons and get_icon tools catch errors and return them as plain text with isError=true. The error messages include raw API errors ('GraphQL error: ...') without actionable next steps for the LLM.
Tool descriptions lack context and recovery hints. 'Search for Font Awesome icons by query string' does not explain when to use search_icons vs get_icon, what format the query should be, or what to do if no results are found. Descriptions should be 50 - 200 characters and include dependency hints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 56 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 48 | - | v1 |
Parameter descriptions are generic and do not explain parameter relationships. The 'style' and 'family' parameters in get_icon have enum constraints but lack guidance on which combinations are valid (e.g., family=sharp does not support style=brands). This forces LLMs to discover valid combinations through trial and error.
No pagination support. The search_icons tool accepts a 'limit' parameter but does not return a next_cursor, total_count, or offset mechanism. If results exceed the limit, the LLM cannot fetch additional pages, it must make multiple search calls with different queries.
Missing chaining IDs in responses. The search_icons tool returns icon names and metadata but does not include all fields needed for downstream get_icon calls. The response should explicitly state which (name, style, family) tuples are needed to retrieve detailed SVG.