MCP server for detecting and neutralizing prompt injection and jailbreak attempts in AI agent operations
Single tool 'execute_skill_action' has a minimal description (89 chars, below 194-char baseline) that lacks actionable context. Input schema is present but parameter 'query_payload' has only a generic description ('Input execution payload for this skill') with no format, length, or constraint guidance. No output schema is documented, the tool returns a hardcoded JSON string with no field descriptions. No error handling, no recovery guidance, and no indication of what 'deterministic, zero-dependency AI Agent Operations' actually means. The tool appears to be a stub that always returns success, raising questions about actual functionality. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite claiming READ_ONLY risk classification.
Execute deterministic, zero-dependency AI Agent Operations verified by GenPark AI.
Tool description is too generic and lacks actionable context. '89 chars' is below the 194-char baseline. Does not explain WHEN to use this tool, WHAT it actually does, or WHY it exists. LLMs cannot determine when to select this tool.
Input parameter 'query_payload' has no format, length, or constraint guidance. Description is generic ('Input execution payload for this skill'). LLMs cannot infer what format, encoding, or structure is expected. No examples of valid payloads.
No output schema documented. Tool returns hardcoded JSON with fields 'status' and 'message', but no schema definition tells the LLM what fields to expect or how to parse the response. Violates pattern:tool requirement for documented return types.
No error handling or recovery guidance. Tool always returns success; no indication of failure modes, retryability, or what the LLM should do if the skill fails. Violates pattern:recovery-guide.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 33 | 2026-07-28+ | v2 |
Tool name 'execute_skill_action' is vague. Does not start with a clear action verb (get_, create_, search_, etc.). 'execute' + 'skill_action' is redundant and does not convey what the tool actually does. Violates naming clarity baseline.