Vibe Coding Discoverrebuff
View on GitHubLLM Prompt Injection Detector
Rebuff is a prompt-injection defense toolkit with heuristic and LLM-based detection, attack-signature storage, and canary-token leak detection. It provides JavaScript/TypeScript and Python SDKs and can use Pinecone or Chroma for vector storage.
Use Cases
Detect prompt injection in user inputDetect canary token leakage in model outputsStore and identify similar attacks using embeddingsAdd layered defenses to LLM applications
Built With
- Language
- TypeScript
- Frameworks
- OpenAI · Pinecone · Chroma · Supabase
Tags
prompt injection · LLM security · detection · canary tokens · attack detection · embeddings · TypeScript SDK · Python SDK