Single tool 'corpmap' has a basic description but critically lacks input schema visibility in the provided source. The tool name is a noun (not verb-prefixed), violating naming conventions. Parameter descriptions exist in the schema specification but the actual tool registration code (corpmap/mcp_server.py) uses a generic build_mcp_server wrapper that obscures schema details. No output schema is documented. The tool appears to perform corporate ownership mapping (a read-only operation), but the implementation details are hidden behind an abstraction layer (cognis_core.mcp.build_mcp_server), making it impossible to verify actual schema compliance from the source provided.
Corporate structure & beneficial-ownership mapper
Tool name 'corpmap' is a noun, not a verb-prefixed action. Should be 'resolve_ownership', 'map_beneficial_owners', or 'get_ownership_structure' to signal intent to LLMs.
Input schema is declared in the evaluation metadata but actual schema registration is hidden in cognis_core.mcp.build_mcp_server(). Cannot verify parameter types, constraints, or descriptions from source code. Tool definition is inferred, not directly visible.
No output schema documented. Tool description does not specify what fields are returned (e.g., owner list structure, effective ownership percentages, entity details). LLMs cannot plan downstream operations without knowing response shape.
Tool description (67 chars) is minimal and lacks context on when to use it vs. alternatives. Does not explain prerequisites (dataset format), error conditions, or what 'beneficial ownership' means in this domain.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
Parameter 'dataset' expects a file path (string) but no validation guidance provided. No mention of required JSON structure, encoding, or error handling if file is missing or malformed.