adk-rag-agent
Build RAG (Retrieval-Augmented Generation) agents with Google ADK and Vertex AI RAG Engine. Use when implementing document Q&A, knowledge base search, or citation-backed responses. Covers VertexAiRagRetrieval tool, corpus setup, and citation formatting.
When & Why to Use This Skill
This Claude skill empowers developers to build sophisticated Retrieval-Augmented Generation (RAG) agents using the Google ADK and Vertex AI RAG Engine. It streamlines the process of connecting AI agents to private document corpora, enabling accurate, citation-backed responses for document Q&A and enterprise knowledge search. By leveraging the VertexAiRagRetrieval tool, it ensures that AI responses are grounded in specific, verifiable data sources, significantly reducing hallucinations in specialized domains.
Use Cases
- Enterprise Knowledge Management: Transform static company wikis, PDFs, and internal documents into an interactive AI assistant that provides instant, cited answers to employee queries.
- Technical Support Automation: Build support agents capable of querying complex product manuals and troubleshooting guides to provide customers with precise instructions and direct links to documentation.
- Legal and Compliance Research: Create specialized agents that can search through vast libraries of regulatory documents or contracts to find specific clauses and provide consolidated summaries with source citations.
- Academic and Research Assistance: Develop tools for researchers to query collections of scientific papers, allowing the AI to synthesize information while maintaining strict attribution to the original authors.
| name | adk-rag-agent |
|---|---|
| description | Build RAG (Retrieval-Augmented Generation) agents with Google ADK and Vertex AI RAG Engine. Use when implementing document Q&A, knowledge base search, or citation-backed responses. Covers VertexAiRagRetrieval tool, corpus setup, and citation formatting. |
Google ADK RAG Agent
Build agents that answer questions from document corpora using Vertex AI RAG Engine.
Requirements
- Vertex AI backend (not Gemini API)
- Google Cloud project with Vertex AI enabled
- RAG corpus created in Vertex AI
Environment Variables
GOOGLE_GENAI_USE_VERTEXAI=1
GOOGLE_CLOUD_PROJECT=your-project-id
GOOGLE_CLOUD_LOCATION=us-central1
RAG_CORPUS=projects/{PROJECT_ID}/locations/{LOCATION}/ragCorpora/{CORPUS_ID}
Core Implementation
from google.adk import Agent
from google.adk.tools import VertexAiRagRetrieval
# Configure RAG retrieval tool
rag_tool = VertexAiRagRetrieval(
name="retrieve_docs",
description="Retrieve relevant documentation for the question",
rag_corpus=os.environ["RAG_CORPUS"],
similarity_top_k=10,
vector_distance_threshold=0.6,
)
# Create agent with RAG tool
agent = Agent(
name="rag_agent",
model="gemini-2.0-flash-001",
instruction=INSTRUCTION_PROMPT,
tools=[rag_tool],
)
Instruction Prompt Pattern
INSTRUCTION_PROMPT = """
You are an AI assistant with access to a specialized document corpus.
RETRIEVAL:
- Use retrieve_docs for specific knowledge questions
- Skip retrieval for casual conversation
- Ask clarifying questions when intent is unclear
SCOPE:
- Only answer questions related to the corpus
- Say "I don't have information about that" for out-of-scope queries
CITATIONS:
- Always cite sources at the end of responses
- Format: [Title](url) or [Document Section](url)
- Consolidate multiple citations from the same source
"""
Corpus Setup
Create corpus via Vertex AI Console or SDK:
from vertexai.preview import rag
# Create corpus
corpus = rag.create_corpus(display_name="my-corpus")
# Import documents (PDF, TXT, HTML)
rag.import_files(
corpus_name=corpus.name,
paths=["gs://bucket/doc.pdf"], # or local files
chunk_size=512,
chunk_overlap=100,
)
Key Parameters
| Parameter | Description | Default |
|---|---|---|
similarity_top_k |
Max chunks to retrieve | 10 |
vector_distance_threshold |
Min similarity (0-1, lower=stricter) | 0.6 |
chunk_size |
Tokens per chunk at import | 512 |
chunk_overlap |
Overlap between chunks | 100 |
Citation Best Practices
- Single source → single citation at end
- Multiple sources → list all citations
- Same document, multiple chunks → consolidate into one citation
- Never expose internal chunk IDs to users