# Agent Sentinel

> The Runtime Authority for AI Agents. Decide what your agents are allowed to do—before they do it.

## What it is

- Budget enforcement that caps spend per action, run, or session to prevent runaway LLM costs.
- Action deny lists that block irreversible or dangerous operations before they execute.
- Approval workflows and human intervention registries that keep a human in the loop.
- An immutable, audit-ready ledger of every decision, input, and policy check for compliance (including EU AI Act Article 14).
- A Python SDK (`agentsentinel-sdk`) that deploys with an agent stack in minutes.

## Docs

- [Agent Sentinel](https://agentsentinel.dev): marketing site and product overview.
- [Documentation](https://docs.agentsentinel.dev): integration guides and API reference.
- [Console](https://console.agentsentinel.dev): policy, budget, and approval management.
- [API](https://api.agentsentinel.dev): the runtime platform the SDK talks to.
- [PyPI package](https://pypi.org/project/agentsentinel-sdk): `pip install agentsentinel-sdk`.

## Contact

- Security: see https://agentsentinel.dev/.well-known/security.txt