Framework-agnostic MCP proxy that aggregates multiple vertical MCP servers (Flight, Hotel, Holiday Package, Vacation Renting) into one namespaced endpoint with conformance validation and tool call logging
Four travel-domain tools with basic schemas and descriptions. All tools have input schemas with typed parameters and descriptions, meeting minimum requirements. However, descriptions are generic (50-80 chars) and lack actionable context for LLM selection. No output schemas documented. Parameter descriptions are minimal, e.g., 'Departure date in YYYY-MM-DD format' lacks guidance on valid ranges, constraints, or error recovery. No error handling guidance. Tools follow verb_noun naming (search_*) but lack composition hints (e.g., which tool to call first, how results chain). No pagination documented despite travel APIs typically returning large result sets. Security and audit patterns absent.
Search for flights from origin to destination on a specific departure date
Search for bundled holiday packages (flight + hotel) to a destination
Search for hotels at a destination for a specific date range
Search for vacation rental properties (apartments, houses) at a destination
Output schemas not documented. LLMs cannot infer what fields to expect from search results, preventing downstream tool chaining and forcing agents to guess at response structure.
Descriptions lack actionable context. Generic 50-80 char descriptions ('Search for flights...') do not explain WHEN to use each tool vs. alternatives, what results look like, or prerequisites. LLMs cannot distinguish between search_flights and search_holiday_packages without deeper reasoning.
No pagination or result limits documented. Travel search APIs typically return 10-100+ results. Without limit/offset parameters and documented caps, agents risk context window exhaustion and degraded reasoning.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2025-06-18+ | v2 |
Parameter descriptions lack constraints and error guidance. 'Departure date in YYYY-MM-DD format' does not specify valid ranges (e.g., must be future date, within 365 days), what happens if invalid, or how to recover. LLMs cannot self-correct without explicit guidance.
No tool composition hints. Agents do not know which tool to call first, how results from one tool feed into another, or whether to call search_flights + search_hotels separately or use search_holiday_packages. Missing dependency documentation.