Generate MCP servers with built-in observability from an OpenAPI spec.
Klaridian is a code generator that emits MCP servers from OpenAPI specs. The 9 tools shown are generated artifacts (Petstore API). Tool naming follows verb_noun convention (addPet, deletePet, findPetsByStatus), which is good. However, descriptions are severely deficient: most are parameter-level descriptions (e.g., 'ID of the order that needs to be deleted') rather than tool-level descriptions explaining WHAT the tool does and WHEN to use it. Per the rubric, tool descriptions must be 10 - 1024 chars and answer what/when/why; these fail that bar. Schemas are present and properly typed (JSON Schema with properties, required fields, enums), but parameter descriptions are missing or minimal. Error handling, output schemas, and composition guidance are not visible in the provided code. The generator itself is sophisticated (OpenAPI→MCP transpilation, observability hooks), but the emitted tool definitions lack LLM-optimized descriptions and parameter documentation.
Create a new pet in the store
Created user object
The JSON request body.
ID of the order that needs to be deleted
Pet id to delete
The name that needs to be deleted
Status values that need to be considered for filter
Tool descriptions are parameter-level, not tool-level. 'ID of the order that needs to be deleted' describes the parameter, not the tool. LLMs cannot determine WHEN to call deleteOrder vs other tools.
getInventory has NO description at all. LLM cannot infer what this tool returns or when to call it.
Parameter descriptions are missing or minimal. E.g., 'requestBody' in addPet has description 'Create a new pet in the store' (tool-level, not param-level). LLMs cannot understand what fields are required or what each field controls.
createUsersWithListInput has ambiguous naming. 'WithListInput' is not a clear action verb. Should be 'createUsersBatch' or 'createMultipleUsers' to clarify it accepts an array.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | 2026-07-28+ | v2 |
Tags to filter by
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields (e.g., does addPet return a pet_id for use in other tools?).