Scoring was not performed

• Document the complete output schema for each tool in its description or in a formal JSON Schema section. For Scan/ScanContent, describe the ScanResult structure (findings array, rules_loaded count, any summary stats). For ListRules, describe RuleInfo fields (id, name, severity, category). For ExplainRule, describe RuleDetail fields. For Discover, describe DiscoverResult and MCPServer fields.
• Replace the generic 'opts' parameter with explicit, typed parameters. For ListRules, add a 'category' parameter (string, optional, enum of valid categories). For Scan/ScanContent, expose key options as explicit parameters: 'workers' (integer, 1 - 16), 'minSeverity' (enum: 'info|low|medium|high|critical'), 'maxFileSize' (integer bytes, optional), 'ignorePatterns' (array of strings, optional). This makes the LLM's intent clear without guessing.
• Add error handling guidance to descriptions. E.g., Scan: 'Scans a file or directory for security issues. Returns findings with severity levels (critical, high, medium, low, info). If the path does not exist or is not readable, returns an error. For permission issues, check file access. For large directories, consider setting maxFileSize or workers options to control resource usage.'
• Document what categories are available for rule filtering. In ListRules, state: 'Returns all rules or filters by category. Valid categories include: [list or enum values from the codebase].'
• Add a 'severity' enum constraint to rule-related responses. When ExplainRule returns a RuleDetail, document that severity is one of: 'critical|high|medium|low|info' so the LLM knows the valid values.