Vibe Coding DiscoverAegis
View on GitHubRuntime policy enforcement for AI agents. Cryptographic audit trail, human-in-the-loop approvals, kill switch. Zero code changes.
AEGIS intercepts AI agent tool calls to enforce runtime policies, block risky actions, request human approval, and record a tamper-evident audit trail. It includes a gateway and monitoring dashboard, with integrations designed to avoid changing existing agent logic.
Use Cases
Block unsafe agent tool calls before executionRequire human approval for sensitive actionsAudit and monitor agent activityEnforce policies across deployed agentsDetect agent security threats
Built With
- Language
- TypeScript
- Frameworks
- LangChain · MCP
Tags
agent security · policy enforcement · audit trail · human approval · tool-call interception · agent monitoring · compliance · kill switch · MCP