Pillar guide10 min read
What Is AI Agent Security? A Practical Guide
Learn the controls AI agents need beyond prompt filtering: identity, least privilege, credential isolation, budgets, revocation, audit trails and human approval.
Read the guidePassControl learning center
Security guidance for agents that actually act.
Plain-language, technically precise guides to identity, credentials, authorization, budgets, revocation and proof. Every guide names the controls PassControl covers—and the ones it does not.
Identity guide10 min read
AI Agent Identity: Authentication and Authority
Learn how to give every AI agent a distinct identity, bind it to an owner, separate authentication from authority, and manage its full security lifecycle.
Read the guideArchitecture guide11 min read
AI Agent Credential Gateways: A Practical Guide
See how an AI agent credential gateway keeps provider keys outside agent runtimes while enforcing identity, scope, budget, revocation and audit controls.
Read the guideEvidence guide10 min read
Verifiable AI Agent Audit Trails and Signed Receipts
Learn how minimal audit events, hash chaining and digital signatures create AI agent call receipts that can be verified without trusting a dashboard export.
Read the guideTooling guide9 min read
MCP Server Security: Controlling Agent Tool Access
How to secure Model Context Protocol servers: treat tool metadata as untrusted input, authenticate the agent, scope each tool, and keep upstream keys out.
Read the guideOperations guide8 min read
AI Agent Budget Controls and Denial-of-Wallet Defence
Denial of wallet is a security failure, not a billing surprise. How to reserve spend before a call, settle it after, and pick the right scope for every limit.
Read the guideOperations guide9 min read
How to Revoke AI Agent Access Quickly and Prove It
Revocation is the control you test, not the one you assume. Layered kill switches, why rotation is not revocation, and how to make a shutdown checkable afterwards.
Read the guideIntegration guide8 min read
GitHub Access for AI Agents Without Sharing Your Token
Let an AI agent read GitHub without holding your token: a fine-grained token as the ceiling, per-agent rules as the floor, and a gateway that injects the key.
Read the guide