A multi-tier application demonstrating user federation through an AI agent system. Includes a secured API, AgentCore runtime with MCP gateway, Flask backend proxy, and React frontend with Okta authentication.
This server exposes 4 tools via AWS Bedrock AgentCore, but critical definition elements are missing or severely underdeveloped. Tool names follow verb_noun convention (get*), which is good. However, descriptions are generic and lack actionable context for LLM selection. Most critically: NO input schemas are visible in the source code for 3 of 4 tools. Only getUserFoodPreferences shows a partial schema (season parameter). Parameter descriptions are absent or trivial. Output schemas are completely undocumented. Error handling guidance is missing. The codebase shows infrastructure (Dockerfile, CDK, Flask backend) but does NOT expose the actual tool definitions, schemas, or implementation details needed for evaluation. Tool definitions appear to be generated or registered dynamically by the Bedrock AgentCore framework, making them impossible to verify from static code analysis.
Retrieve user identity, account, or personal information
Retrieve user's past interactions or conversation history
Retrieve user's food or dining preferences with optional season filter
Retrieve complete user profile data including identity, account, conversations, and food preferences
No input schemas visible for 3 of 4 tools (getUserProfileComplete, getUserBasicInfo, getUserConversations). Only getUserFoodPreferences shows a partial schema with one parameter.
Descriptions are generic and under 100 characters. E.g. 'Retrieve complete user profile data...' lacks WHEN to use this vs getUserBasicInfo, WHAT fields are returned, or any prerequisites.
No output schemas documented. LLMs cannot plan downstream calls or extract required fields. E.g. does getUserProfileComplete return user_id, account_id, conversation_ids? Unknown.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 37 | 2026-07-28+ | v2 |
Tool definitions are not explicitly visible in source code. They appear to be registered dynamically by bedrock-agentcore framework. Cannot verify actual schemas, parameter types, or error handling.
No error handling guidance. If a user is not found, what does the tool return? Can the agent retry? Should it search for alternatives? No recovery path documented.