Industrialized FastMCP 3.2.0 compliant control plane for VRChat. Provides unified Portmanteau tools for avatar, OSC, and system orchestration with avatar state tracking, parameter management, avatar switching, NPC behaviors, and OSC communication.
Six tools with significant structural issues. All tools use a 'portmanteau' pattern (single tool with operation parameter) rather than verb_noun naming, violating the core naming convention. Descriptions are present but generic and lack actionable guidance. Schemas are visible but incomplete: parameters lack type constraints, enums are missing for operation fields, and output schemas are undocumented. No error handling guidance. The manage_avatar tool shows partial implementation in source (AvatarManager class exists) but tool registration is not visible in provided code, making actual schema verification impossible. manage_social combines 6 distinct operations (list_friends, friend_status, search_users, get_user_info, list_notifications, respond_notification) in one tool, violating single-responsibility principle.
Unified Portmanteau for Avatar State and Parameter Management. Operations: get_state (Retrieve full state OSC + REST metadata), load (Trigger avatar load by ID), set_param (Set parameter value supports interpolation), get_param (Retrieve specific parameter value).
Unified Portmanteau for VRChat Creator Economy (REST API REQUIRED). Operations: balance (Retrieve current VRChat Credit balance), products (List active Udon products and subscriptions).
Unified Portmanteau for OSC Protocol Interaction. Operations: send (Dispatch OSC message to specific address with args), stats (Retrieve real-time protocol traffic statistics).
Unified Portmanteau for Friends, Users, and Notifications (REST API REQUIRED). Operations: list_friends (List friends, offline=True for offline friends instead of online/active), friend_status (Check friendship/request status with a specific user_id), search_users (Search for users by username or query), get_user_info (Get detailed user profile information), list_notifications (List pending notifications), respond_notification (Accept/decline friend requests or invitations).
Unified Portmanteau for System Telemetry and Diagnostics. Operations: health (Get server health status and uptime), metrics (Retrieve performance metrics including request count, error rate, response times), config (Get current server configuration), logs (Retrieve recent server logs).
All six tools use 'manage_*' + operation parameter pattern instead of verb_noun naming (e.g., get_avatar_state, load_avatar, set_avatar_parameter). LLMs cannot infer intent from 'manage_avatar' alone and must parse the operation enum, increasing reasoning overhead and error likelihood.
manage_social combines six distinct operations (list_friends, friend_status, search_users, get_user_info, list_notifications, respond_notification) in a single tool. Violates single-responsibility principle. Should be split into get_friends, get_user_info, search_users, get_notifications, respond_to_notification.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., user_id, channel_id for chaining). Responses are unstructured.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | <=2025-11-25 | v2 |
Unified Portmanteau for World Discovery and Instance Telemetry (REST API REQUIRED). Operations: get_info (Fetch metadata for a world ID), search (Search for active worlds by query string).
Operation parameters lack enum constraints. 'operation' field accepts free-form strings (get_state, load, set_param, get_param) without formal enum declaration. LLMs may hallucinate invalid operations.
No error handling guidance. Tools do not document what errors are retryable, user-fixable, or fatal. No recovery suggestions (e.g., 'If avatar not found, call search_avatars first').