/add-streaming

Arguments: $ARGUMENTS

Goal: Add bidirectional streaming, real-time updates, and streaming capabilities to Google ADK agents for interactive, low-latency applications.

Core Principles:

• Understand existing agent structure before adding streaming
• Follow Google ADK streaming patterns and best practices
• Implement proper error handling and connection management
• Provide real-time feedback during implementation

Phase 1: Discovery Goal: Understand the target agent and streaming requirements

Actions:

• Parse $ARGUMENTS to identify target agent or feature
• If $ARGUMENTS is unclear, use AskUserQuestion to gather:
  • Which agent needs streaming capabilities?
  • What type of streaming is needed (bidi, server-to-client, client-to-server)?
  • What real-time features are required?
  • Any specific performance or latency requirements?
• Detect project structure and Google ADK version
• Example: !{bash find . -name "package.json" -o -name "pyproject.toml" | head -5}

Phase 2: Analysis Goal: Understand existing agent implementation and identify integration points

Actions:

• Locate target agent files using Glob
• Example: !{bash find . -type f -name "agent.py" -o -name "agent.ts" | head -10}
• Read agent configuration and implementation files
• Identify current communication patterns
• Check for existing streaming infrastructure
• Understand dependencies and requirements

Phase 3: Planning Goal: Design the streaming implementation approach

Actions:

• Determine streaming architecture based on use case:
  • Bidirectional streaming for interactive conversations
  • Server-to-client for real-time updates
  • Client-to-server for continuous input processing
• Identify required dependencies (grpc, websockets, etc.)
• Plan error handling and reconnection logic
• Present implementation plan to user for confirmation

Phase 4: Implementation Goal: Add streaming capabilities to the agent

Actions:

Task(description="Add streaming capabilities", subagent_type="google-adk-streaming-specialist", prompt="You are the google-adk-streaming-specialist agent. Add bidirectional streaming and real-time features to $ARGUMENTS.

Context: Google ADK agent requiring streaming capabilities

Requirements:

• Implement bidirectional streaming using Google ADK patterns
• Add proper connection lifecycle management (connect, disconnect, reconnect)
• Implement error handling and retry logic
• Add real-time event handling and callbacks
• Follow Google ADK best practices for streaming
• Include code comments explaining streaming flow
• Handle backpressure and flow control
• Implement graceful degradation on connection issues

Streaming Patterns to Consider:

• gRPC bidirectional streaming for agent-to-agent communication
• WebSocket connections for browser-based real-time updates
• Server-Sent Events (SSE) for one-way server-to-client streaming
• Streaming response handling with chunked data processing

Expected output:

• Updated agent files with streaming capabilities
• Connection management implementation
• Error handling and recovery logic
• Example usage showing streaming in action
• Documentation of streaming API")

Phase 5: Verification Goal: Verify streaming implementation works correctly

Actions:

• Check that streaming code follows Google ADK patterns
• Verify error handling and connection management
• Review example usage and documentation
• If tests exist, run them: !{bash npm test || python -m pytest || echo "No tests configured"}
• Validate code quality: !{bash npm run lint || python -m pylint . || echo "No linter configured"}

Phase 6: Summary Goal: Document what was accomplished

Actions:

• Summarize streaming features added:
  • Type of streaming implemented
  • Connection management approach
  • Error handling strategy
  • Key files modified
• Highlight streaming API usage
• Suggest next steps:
  • Test with real-time data
  • Monitor connection stability
  • Optimize for latency
  • Add metrics and monitoring