Loading ToolBench
Agentic lead scoring and conversion velocity evaluation server that calculates behavioral engagement scores and predicts high-intent checkout triggers for marketing automation.
Single tool 'evaluate_lead_score' has a basic description (71 chars) but critical gaps in schema documentation and parameter descriptions. Input schema is visible in the evaluation request but NOT in the source code itself, the mcp_server.py tools/list response omits the schema entirely, showing only name and description. This violates MCP spec requirements for tool registration. The client.py method signature reveals parameters, but the MCP server does not expose them in the protocol response. Parameter descriptions are absent from the schema. No output schema is documented. The tool performs a deterministic calculation with no error handling, recovery guidance, or validation of input ranges.
Scores customer engagement velocity and outputs next best marketing action.
Tool schema not exposed in MCP protocol response. tools/list returns only name and description; input schema is missing from the JSON-RPC response.
No parameter descriptions in schema. Parameters (session_dwell_time_sec, page_views_count, cart_abandoned_previously, coupon_applied) lack descriptions explaining their meaning, valid ranges, or constraints.
Output schema not documented. The tool returns a complex object with fields like 'total_intent_score', 'funnel_stage', 'next_best_action', but no schema definition is provided to the LLM.
No input validation or error handling. Tool accepts any integer for session_dwell_time_sec and page_views_count without bounds checking. No error messages guide recovery.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
Tool name 'evaluate_lead_score' is vague. Does not start with a clear action verb. 'score_lead_velocity' or 'predict_conversion_stage' would be more explicit about the action.