MCP server providing trading and broker operations access through the LumosTrade library
LumosTradeTool has 2 tools with critical definition gaps. Both tools have names starting with action verbs (positive), but descriptions are extremely minimal, parameter descriptions are absent, input schemas lack detail, and output schemas are undocumented. The tool definitions appear inferred from code rather than explicitly registered with full metadata. Critical missing elements: no parameter descriptions, no output schema documentation, no error handling guidance. Tool names are clear but generic descriptions and undocumented schemas severely limit LLM reasoning and agent composition.
Retrieve stock or option quotes for specified symbols
Refresh broker connections and authentication tokens
Tool descriptions are too brief (under 20 characters). 'Retrieve stock or option quotes for specified symbols' is adequate but 'Refresh broker connections and authentication tokens' lacks clarity on when to use this vs alternatives, prerequisites, or side effects.
Parameter 'symbols' in getQuotes has no description. LLMs cannot determine expected format (ticker symbols? option symbols? comma-separated? array?), valid length/range, or required format constraints. Violates pattern:tool-description requirement that every parameter must have a non-empty description.
refreshBrokers has NO input schema documented. Missing schema prevents validation and leaves LLMs guessing about parameters, authentication requirements, or side effects. A WRITE-risk tool requires explicit input schema and constraints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 40 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 15 | - | v1 |
No output schemas documented for either tool. LLMs cannot plan downstream calls without knowing what fields are returned. 'Retrieve stock or option quotes' does not specify: are results structured objects? Arrays? What fields (symbol, price, bid, ask, timestamp, etc.)? This forces agents to guess and invites errors.
No error handling guidance. What happens if a symbol is invalid? If broker authentication fails? If rate limits are hit? Responses must tell LLMs what to do next (retry, ask user, try alternative, etc.). Currently absent.
refreshBrokers is marked WRITE-risk but has no confirmation request or dry-run option. Agents may inadvertently re-authenticate, disrupting active sessions. Destructive/side-effect tools should support confirmation patterns.
Tool definitions appear inferred from filenames and brief descriptions rather than explicitly registered with full schema metadata. Source code shows tool references ('LumosAI/Tools/LumosTradeTool/src/tools/getQuotes.ts') but actual tool registration code with complete schemas is not visible in provided source.