DriftCop presents a severe case of inferred tool definitions with minimal visible schema validation and almost no parameter descriptions. The tool definitions appear only in example file paths (javascript_tool.js, python_tool.py) with no evidence of actual implementation, registration, or runtime behavior in the server codebase. The source shows a Makefile, pyproject.toml, and package.json for a security scanner project, but NO MCP server implementation code is provided. Tool names are present but lack proper verb-noun conventions (execute_command, query_database, call_api are overly generic). None of the 7 tools have visible descriptions in the source code, the descriptions provided in the evaluation prompt appear to be inferred or aspirational. Input schemas are partially visible in the prompt but lack parameter descriptions and have inconsistent constraint application (e.g., delete_file has additionalProperties:true despite being marked DESTRUCTIVE, undermining input validation). No output schemas are documented. Error handling is not visible. Security concerns are severe: execute_command and delete_file are dangerously exposed without confirmation or permission gates.

• ADD PARAMETER DESCRIPTIONS TO ALL TOOLS. For each parameter, explain its purpose, valid format, and constraints. E.g., 'pattern (string): Glob or regex pattern to match files. Use * for wildcards, e.g., '*.py' for Python files.'
• DOCUMENT OUTPUT SCHEMAS FOR ALL TOOLS. E.g., search_files should return: {type: object, properties: {files: {type: array, items: {type: string}}, total: {type: integer}, next_cursor: {type: string or null}}}
• RENAME GENERIC TOOLS TO BE SPECIFIC. Rename execute_command to run_shell_command. Rename call_api to make_http_request. Rename query_database to execute_sql_query. Include the resource type in the name so the LLM can distinguish it from others.
• ADD SECURITY GATES TO DESTRUCTIVE TOOLS. Implement permission checks for execute_command, delete_file, write_file. Require confirmation parameter and validate it explicitly. Return clear error: 'Delete not confirmed. Pass confirm=true only if you understand this action is irreversible.'
• VALIDATE ALL INPUT. Set bounds on arrays (e.g., max 100 items), string lengths (max 4096 chars), and enforce enums for categorical params. Reject invalid input with messages like: 'Invalid method: got POST, expected one of: GET, POST, PUT, DELETE'.
• ADD DRY-RUN SUPPORT TO WRITE/DELETE TOOLS. Add a 'dry_run=true' parameter to execute_command, write_file, delete_file. Return what would happen without actually executing. E.g., 'Would delete: /etc/shadow. Are you sure? (This action is IRREVERSIBLE.)'
• DOCUMENT ERROR SCENARIOS AND RECOVERY. E.g., 'If file not found, try search_files() to locate it first. If permission denied, check that the agent has read/write scope for this path.'