Loading ToolBench
Scores company eligibility for non-dilutive capital and credit lines based on monthly spend and LTV/CAC ratio metrics
This MCP server exhibits critical definition quality gaps. The single tool 'score_eligibility' has a reasonable description but lacks proper parameter descriptions, has minimal schema rigor, and provides no error handling guidance. The tool performs a simple eligibility check with hardcoded thresholds, but the implementation is not presented as a proper MCP server, only a Python client class. No MCP protocol scaffolding, tool registration, or server implementation is visible in the source. The code suggests this is a bare Python class, not an actual MCP server conforming to the protocol specification.
Scores company eligibility for non-dilutive capital based on monthly spend and LTV/CAC ratio, returning eligibility status, credit score, and term sheet readiness
Tool description lacks actionable detail and WHEN to use it. Current description states WHAT it returns but does not explain eligibility thresholds, prerequisites, or how the LLM should interpret the credit_score field or term_sheet_ready flag.
Input parameters 'monthly_spend' and 'ltv_cac_ratio' lack descriptions entirely in the schema. LLMs cannot infer what values are acceptable or what ranges are valid (e.g., is $0 valid? Is ltv_cac_ratio capped at 10?).
Output schema is not documented. Tool returns a dict with three keys (eligibility_status, credit_score, term_sheet_ready) but the response structure, field types, and value ranges are not declared. LLMs cannot reliably parse or validate outputs without a documented schema.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 35 | 2026-07-28+ | v2 |
No error handling or recovery guidance. If monthly_spend is negative or ltv_cac_ratio is 0, the tool silently returns 'NURTURE'. LLMs receive no guidance on what went wrong, whether to retry, or how to fix the input.
Magic thresholds hardcoded in implementation (monthly_spend >= 50000.0, ltv_cac_ratio >= 2.0, credit_score = 845 if ready else 620). These are not configurable and not explained in the tool description. Users/agents have no way to understand why $49,999.99 is rejected or how to improve a score of 620.
No MCP server implementation visible. The source code shows only a Python client class and example usage, not an actual MCP server with tool registration, message handling, or protocol compliance. It is unclear whether this is intended to be an MCP server at all.