bumpers

guardrails for ai agents

an API and SDK for building safe and reliable AI agents that stay on task and within bounds

example.py

Key Features

Action Validation

Validate and restrict AI agent actions using customizable rules and policies.

Content Filtering

Filter and sanitize content generated by AI agents to ensure safety and compliance.

Semantic Drift Detection

Monitor and prevent AI agents from deviating from their intended purpose.

Real-time Monitoring

Track and analyze AI agent behavior with comprehensive logging and analytics.

Policy Management

Define and manage safety policies using simple YAML configuration files.

Framework Agnostic

Works with any AI framework or LLM, providing consistent safety guardrails across your stack.

LangChain

OpenAI

Anthropic

LlamaIndex

Mistral