This is a Kotlin static analysis framework (Detekt rules + policy checker), not an MCP server. The three tools (readText, writeText, env) are defined in Kotlin source code, not as MCP tool registrations. No MCP protocol implementation is visible. Tool definitions lack proper JSON Schema structures, parameter type declarations are minimal, and descriptions are extremely brief (10-20 chars). No error handling, no output schemas, no pagination. The codebase enforces tool policy via compile-time checks (ToolBodyForbiddenApis, ToolPolicyCapabilityComparator) but does not expose an MCP interface.

• Implement an actual MCP server (HTTP/SSE or Streamable HTTP transport) that registers these three tools with proper MCP protocol messages. The current codebase is a compile-time policy checker, not a runtime agent tool.
• Expand tool descriptions to 50-200 characters. Example: 'readText: Read file contents within policy-allowed paths. Returns file text or error if path violates read globs. Use to fetch configuration, logs, or source code.'
• Define complete JSON Schema for each tool's input. readText should specify: path (string, required, pattern for allowed globs), optional encoding (enum: utf-8, ascii, iso-8859-1). writeText should add: content (string, required), overwrite (boolean, default false).
• Document output schemas. readText returns {content: string, path: string, size_bytes: number}. writeText returns {path: string, bytes_written: number, success: boolean}. env returns {name: string, value: string}.
• Add error handling with recovery hints. E.g., 'Path /etc/passwd violates read policy. Allowed globs: /home/*, /tmp/*. Try a path matching one of these patterns.' Include error codes (POLICY_VIOLATION, NOT_FOUND, PERMISSION_DENIED).