Sandbox Resources Discovery Skill

Purpose

When you run npx ampx sandbox, it deploys AWS resources (DynamoDB tables, Lambda functions, etc.) to your AWS account. This skill helps you quickly discover what resources are deployed without manually checking the AWS console.

Common questions this skill answers:

• "What DynamoDB tables are in my sandbox?"
• "What Lambda functions did my sandbox create?"
• "What's the physical AWS name of the Company table?"
• "Show me my sandbox resources"

Activation

Invoke with simple phrases like:

• "Discover my sandbox"
• "Show my sandbox resources"
• "What's in my sandbox?"
• "List sandbox tables and functions"
• "Sandbox discovery"

Example:

User: "Discover my sandbox"

Claude: Uses MCP to find all DynamoDB tables and Lambda functions,
then shows you what exists with physical AWS names.

How It Works

When you ask about your sandbox resources:

1. Reads amplify_outputs.json from your Amplify project (local file, no AWS API needed yet)
2. Queries AppSync API using the GraphQL URL from amplify_outputs.json
3. Discovers all resources connected to AppSync: DynamoDB tables, Lambda functions, data models, auth config
4. Presents complete inventory with physical AWS names, schemas, and metadata you can use with AWS CLI

What Gets Discovered

DynamoDB Tables

• Table name (physical AWS name)
• Data source name (how AppSync references it)
• Region
• Full table ARN for CLI access

Example:

ClaudeQueryLog Table
Physical name: ClaudeQueryLog-4mogzwizrffkrmux5kud2mtuiy-NONE
Region: us-east-1

Lambda Functions

• Function name
• Function ARN (for invocation)
• Data source name (how AppSync references it)
• Runtime and configuration

Example:

queryClaude Function
ARN: arn:aws:lambda:us-east-1:755956835466:function:queryClaude
Data Source: QueryClaudeLambdaDataSource

Data Models (GraphQL Schema)

• Model name (Conversation, Message, ClaudeQueryLog, etc.)
• Field count
• Field definitions
• Authorization rules
• Custom indexes

Example:

Conversation Model
Fields: 5 (id, topic, createdAt, updatedAt, messages)
Authorization: Public API Key access
Custom Index: conversationsByCreatedAt

Authentication Configuration

• Cognito User Pool ID
• Identity Pool ID
• OAuth providers (Google, etc.)
• OAuth domain
• Password policy

Workflow

Step 1: Discover Resources

When user asks "discover my sandbox", Claude:

1. Calls the MCP tool amplify_discover_sandbox_resources()
2. Tool reads amplify_outputs.json from project directory
3. Tool queries AppSync using GraphQL URL from amplify_outputs.json
4. Returns complete inventory: tables, functions, models, auth config

Step 2: Display Results

Claude presents formatted output:

🔌 AppSync API: 4mogzwizrffkrmux5kud2mtuiy
   URL: https://rg7i3n4o65cgzgquwyamxpki44.appsync-api.us-east-1.amazonaws.com/graphql

📊 DynamoDB Tables (3 found):
• ClaudeQueryLog-4mogzwizrffkrmux5kud2mtuiy-NONE
• Conversation-4mogzwizrffkrmux5kud2mtuiy-NONE
• Message-4mogzwizrffkrmux5kud2mtuiy-NONE

⚡ Lambda Functions (1 found):
• queryClaude (arn:aws:lambda:us-east-1:755956835466:function:queryClaude)

📚 Data Models (3 found):
• Conversation (5 fields, public access)
• Message (9 fields, public access)
• ClaudeQueryLog (10 fields, public access)

🔐 Authentication:
• User Pool: us-east-1_JAZUgr7FJ
• OAuth: Google

Step 3: Enable Further Action

User can then:

• "Query the Company table with AWS CLI"
• "Show me what's in the User table"
• "Invoke the databricksAgent function"
• "Check Lambda function logs"

Critical Rules

1. Always discover first - Get real resource names before suggesting CLI commands
2. Show physical names - Users need the full amplify-... names for AWS CLI
3. Provide copy-paste CLI - Make it easy to run commands immediately
4. Handle errors gracefully - If sandbox isn't running, explain clearly
5. Region awareness - Default to us-east-1 but allow override

Common Use Cases

"I just started my sandbox - what got created?"

→ Discover resources, show all tables and functions

"How many items are in each table?"

→ Show item counts for all DynamoDB tables

"I want to query the TransportationInsight table"

→ Discover, find physical table name, provide AWS CLI command

"What Lambda functions do I have?"

→ List all functions with runtime and timeout details

"Check if my data was created"

→ Show table item counts to verify data exists

Error Handling

If amplify_outputs.json is missing:

Error: amplify_outputs.json not found in current directory

Make sure you've run: npx ampx sandbox

Try these steps:
1. Navigate to your Amplify project: cd my-amplify-project
2. Start the sandbox: npx ampx sandbox
3. Wait 2-3 minutes for resources to deploy
4. Then ask "discover my sandbox"

If sandbox AppSync API not found:

Error: Could not find AppSync API matching URL

The sandbox may not be running. Run "npx ampx sandbox" to start it.

If AWS credentials are wrong:

Error: Failed to query AppSync APIs - Access denied

Verify:
1. AWS_ACCESS_KEY_ID is set
2. AWS_SECRET_ACCESS_KEY is set
3. User has AppSync permissions (appsync:ListGraphqlApis, appsync:ListDataSources)

Tips & Tricks

1. Check Sandbox Health

User: "What resources are in my sandbox?"
Claude: Discovers and shows counts
Result: Verify everything deployed correctly

2. Monitor Data Creation

User: First ask "discover sandbox"
User: Run your test
User: Ask "discover sandbox again" - compare item counts
Result: See exactly how many records were created/deleted

3. Find Physical Names for CLI

User: "Discover my sandbox"
Claude: Shows physical names like "amplify-tiapp-feather-san-company-abc123def"
User: Copy/paste that name into AWS CLI commands

4. Multi-Region Testing

User: "Show my sandbox in us-west-2"
Result: Discover resources in different region

Related Skills & Tools

• AWS CLI - Use the physical names discovered here to query tables manually
• handbook - Learn best practices for data modeling in Amplify
• webapp-testing - Test your frontend against the sandbox

See Also

• AWS DynamoDB CLI Reference
• AWS Lambda CLI Reference
• Amplify Data Documentation

Success Criteria

✅ User asks about sandbox resources (e.g., "Discover my sandbox") ✅ Claude calls amplify_discover_sandbox_resources tool ✅ Tool reads amplify_outputs.json from current directory ✅ Tool queries AppSync and discovers all resources ✅ Claude displays: DynamoDB tables, Lambda functions, data models, auth config ✅ User gets physical AWS resource names and ARNs for CLI access ✅ Error messages are clear and guide user to next steps ✅ Works automatically from project directory with amplify_outputs.json present