Expose tools from local MCP servers via NATS, and start an MCP server that makes NATS-exposed tools available as MCP tools
This server has severe definition quality issues. Only 2 tools are present, both with critical gaps. hello_echo has a basic but acceptable schema and minimal description. query_order_data has a completely undescribed parameter ('question' has description: null) and an overly long, vague description that reads like a sales pitch rather than a precise tool specification. Neither tool has any error handling guidance, output schemas are not documented, and there is no evidence of schema validation or recovery hints. The server appears to be a proof-of-concept for exposing MCP tools via NATS pub/sub, but the tools themselves are not production-ready.
Say hello to someone
This tool can be used to query order data. The users question is used as a plain input into this tool. It can answer all queries about order ids, delivery/shipment locations, order value, total order value for a given timeframe, how much revenue in a certain timeframe etc.
query_order_data has null parameter description: the 'question' parameter has no description (null in schema). LLMs cannot infer meaning from the parameter name alone.
query_order_data description is 252 characters and reads as a marketing statement ('This tool can be used to...can answer all queries...etc.') rather than a precise, LLM-optimized specification. Rubric baseline for tool descriptions is 194 chars (p10=34, p90=392). The description should state WHAT the tool does, WHEN to use it, and WHAT it returns, not what it 'can be used to' do.
No output schemas documented for either tool. LLMs need to know what fields to expect so they can plan downstream tool calls and extract the right data.' Without output documentation, agents cannot plan multi-step workflows.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 39 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 31 | - | v1 |
No error handling guidance. Neither tool's description explains what errors are possible, how to recover, or what the LLM should do if the call fails.
hello_echo description is minimal (21 chars: 'Say hello to someone'), at the lower bound of the acceptable range. It lacks context about when to use this tool vs other communication mechanisms, or what the return value represents.
query_order_data parameter 'question' accepts a string with no constraints (no enum, no format, no min/max length).
query_order_data parameter is marked required=false but the description does not explain what happens if it is omitted.