Knowledge Base¶

Automatic RAG with Amazon Bedrock Knowledge Bases. Seamless memory across sessions.

How It Works¶

flowchart TD
    A["User Query"] --> B["1. RETRIEVE<br/>Search KB for relevant context"]
    B --> C["2. AGENT EXECUTION<br/>Process query with KB context"]
    C --> D["3. STORE<br/>Save conversation to KB"]
    D --> E["Future queries benefit<br/>from stored knowledge"]

When DEVDUCK_KNOWLEDGE_BASE_ID is set, DevDuck automatically:

Before each query — Retrieves relevant context from the knowledge base
Runs the agent — Processes query enriched with retrieved context
After each response — Stores the conversation for future reference

Quick Start¶

One environment variable:

export DEVDUCK_KNOWLEDGE_BASE_ID="your-kb-id-here"
devduck

Every query now has automatic RAG — no manual tool calls needed.

Manual Tool Usage¶

You can also use the knowledge base tools directly:

# Retrieve from knowledge base
retrieve(
    text="how do I configure bedrock?",
    knowledgeBaseId="your-kb-id"
)

# Store to knowledge base
store_in_kb(
    content="Important finding: the API rate limit is 1000/min",
    title="API Rate Limits",
    knowledge_base_id="your-kb-id"
)

Configuration¶

Variable	Description
`DEVDUCK_KNOWLEDGE_BASE_ID`	Bedrock Knowledge Base ID for automatic RAG
`STRANDS_KNOWLEDGE_BASE_ID`	Alternative KB ID for the `retrieve` tool

What Gets Stored¶

Each conversation is stored as:

Content: Input: {query}, Result: {response}
Title: DevDuck: {date} | {query preview}

This builds a growing knowledge base that makes future queries more informed.

Cross-Session Memory

Knowledge persists across DevDuck sessions. Start a new session and your previous conversations are available as context.