Quick Start¶

Index Documents¶

Point RavenRustRAG at a directory or file to index:

# Index a directory of documents
ravenrag index ./docs/

# Index a single file
ravenrag index README.md

# Index with verbose output
ravenrag index ./notes/ -v

Supported formats: .txt, .md, .csv, .json, .jsonl, .html

Query Your Documents¶

# Simple semantic search
ravenrag query "how does authentication work"

# Get more results
ravenrag query "error handling patterns" --top-k 10

# Hybrid search (BM25 + vector)
ravenrag query "configuration file" --hybrid

Generate LLM Prompts¶

The prompt command wraps search results in an LLM-ready format:

ravenrag prompt "explain the build process"

Output:

Use the following context to answer the question.

Context:
[1] (score: 0.89) The build process uses cargo build...
[2] (score: 0.85) Release builds are created with...

Question: explain the build process

Check Index Status¶

ravenrag info

Export and Import¶

# Backup your index to JSONL
ravenrag export > backup.jsonl

# Restore from backup
ravenrag import backup.jsonl

Start the HTTP Server¶

# Start on default port 8484
ravenrag serve

# With API key authentication
RAVEN_API_KEY=my-secret-key ravenrag serve --port 8484

Run Diagnostics¶

ravenrag doctor

This checks Ollama connectivity, embedding model availability, and database health.