What It Is
Most AI demos are chatbots — they respond with text. This project implements a true agent loop: Claude doesn’t just answer, it decides which Redis operation to run, executes it, reads the result, and then explains what happened. The key distinction is tool use with feedback.
You: "Set user:1001 to 'Ajay' and expire it in 300 seconds"
↓
AgentLoop → Claude API → Claude calls: set_key tool
↓
RedisTools → jedis.setex("user:1001", 300, "Ajay")
↓
Tool result fed back to Claude
↓
Agent: "Done! Stored 'Ajay' under user:1001 with a 5-minute TTL."
Features
Agent Loop
Full multi-turn loop: user input → Claude API call with tool definitions → tool execution → result back to Claude → final response. Handles multi-step operations in one conversation turn.
Redis Tool Set
Four Redis operations exposed as Claude tools: get key, set key (with optional TTL), delete key, and list all keys. Claude decides which to call based on the user's natural language input.
Clean Architecture
Four focused classes: Main (REPL loop), AgentLoop (orchestration), ClaudeClient (HTTP + tool schema), RedisTools (Jedis operations). Easy to extend with new tools.
Zero Framework
Raw HTTP calls to the Anthropic API — no LangChain, no Spring AI. The agent loop logic is fully visible and easy to study, which is the point of the project.
Stack
language
Java 17+
build
Maven
ai
Claude API (tool use)
redis client
Jedis
redis server
Docker (redis:7-alpine)
Quick Start
# 1. Start Redis
docker run -d --name redis-agent -p 6379:6379 redis:7-alpine
# 2. Set API key
export ANTHROPIC_API_KEY=sk-ant-...
# 3. Build and run
git clone https://github.com/Ajay3007/redis-ai-agent.git
cd redis-ai-agent
mvn clean package -q
java -jar target/redis-ai-agent-1.0-SNAPSHOT.jar
> Set user:42 to "hello" with 60s TTL
Agent: Done — stored "hello" at user:42, expires in 60 seconds.