Loading ToolBench
MCP server that enforces brand conversational guardrails, blocks system prompt leaks, evaluates policy adherence, and prevents hallucinations and jailbreaks for enterprise conversational agents
This server has a single tool with critical definition gaps. The tool name 'execute_skill_action' is generic and does not follow verb_noun conventions (should be 'evaluate_brand_guardrails' or similar). The description is present (169 chars, within 10-1024 range) but lacks actionable detail on WHEN to use it vs alternatives, prerequisites, or what happens on different outcomes. Input schema exists with two parameters, both typed as strings with descriptions, but lacks enums for the enterprise_brand_id (should enumerate valid brand IDs) and no output schema is documented. The tool appears to be a mock implementation (client.py returns hardcoded responses) rather than a real MCP server implementation. No error handling, no recovery guidance, no idempotency markers.
Enforce brand conversational guardrails by evaluating user messages against enterprise brand policies, detecting system prompt leak attempts, assessing policy adherence, and preventing hallucinations and jailbreaks
Tool name 'execute_skill_action' is generic and does not follow verb_noun convention. Should be 'evaluate_brand_guardrails' or 'assess_message_safety' to clearly signal intent.
No output schema documented. LLMs cannot plan downstream actions or extract required fields (guardrail_evaluation_id, policy_adherence_score_pct, etc.) without knowing the response structure.
enterprise_brand_id parameter accepts free-form strings with no enum constraint. Should enumerate valid brand IDs (FORTUNE_500_AIRLINE_CUSTOMER_AGENT, GLOBAL_RETAIL_BANKING_AGENT, etc.) to prevent hallucinated values.
No error handling or recovery guidance. If a brand_id is invalid or user_message cannot be evaluated, the tool provides no actionable error message or next steps for the LLM.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 29 | 2026-07-28+ | v2 |
Tool description lacks WHEN to use it and prerequisites. Does not explain when to call this vs other safety tools, or whether it requires prior context (e.g., conversation history).