MCP server for managing Taikun cloud infrastructure, including projects, virtual clusters, and application catalogs
Scoring was not performed
Destructive operations (delete_project, delete_virtual_cluster) lack confirmation or dry-run support. No description explicitly states that deletion is irreversible, increasing risk of accidental agent-initiated data loss.
Output schemas are not documented in tool definitions. Tools return text responses via createJSONResponse(), but the response structure (ProjectListResponse, ProjectSummary fields) is not declared in the tool registration. LLMs cannot predict response fields for downstream planning.
Descriptions for destructive tools (delete_project, delete_virtual_cluster) do not explicitly state that the operation is irreversible or have side effects. Description is only 30-33 chars; rubric requires 10-1024 chars with state-mutation clarity.
No error handling guidance visible in tool definitions. If an invalid cloudCredentialID or kubernetesProfileID is passed to create_project, the error response does not guide the LLM to discover valid IDs (e.g., 'Call list_credentials() first').
Parameters like cloudCredentialID, kubernetesProfileID, and alertingProfileID accept only numeric IDs. No guidance provided for discovering valid IDs, the tool description does not mention prerequisites (e.g., 'Use list_credentials() to find valid IDs').
Response field naming is not visible in tool definitions. If list_projects returns 'ID' but delete_project expects 'projectID', the mismatch will cause LLMs to fail. Naming conventions must be enforced (e.g., consistently use projectId or project_id across all tools).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 0 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 42 | - | v1 |