{
  "context": "Tests whether the agent can apply comprehensive MCP tool guidelines by identifying and fixing violations in poorly documented tools. Covers naming, parameters, documentation, consistency, side effects, and failure modes.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Naming convention violations",
      "description": "Identifies and fixes naming violations (wrong case, missing prefixes, non-action-oriented names)",
      "max_score": 14
    },
    {
      "name": "Parameter type issues",
      "description": "Fixes language-specific type names, adds missing generic types",
      "max_score": 14
    },
    {
      "name": "Missing constraints/examples",
      "description": "Adds missing parameter constraints, examples, and validation rules",
      "max_score": 14
    },
    {
      "name": "Implementation details in docs",
      "description": "Removes implementation details from tool descriptions",
      "max_score": 9
    },
    {
      "name": "Error documentation",
      "description": "Adds or improves error documentation with consistent formatting",
      "max_score": 9
    },
    {
      "name": "Documentation structure",
      "description": "Ensures consistent documentation structure across tools",
      "max_score": 9
    },
    {
      "name": "Cross-tool consistency",
      "description": "Ensures parameter naming and ordering consistency across tools",
      "max_score": 9
    },
    {
      "name": "Tool organization",
      "description": "Improves tool grouping and discoverability through prefixes and references",
      "max_score": 9
    },
    {
      "name": "Language-agnostic practices",
      "description": "Applies language-agnostic documentation practices throughout",
      "max_score": 5
    },
    {
      "name": "Side effects documentation",
      "description": "Documents side effects, required auth/tenant scope, and idempotency key support where applicable",
      "max_score": 4
    },
    {
      "name": "Failure mode details",
      "description": "Includes failure modes with short 'when/why' notes for retry/fallback guidance",
      "max_score": 4
    }
  ]
}