FAQ & troubleshooting

Does agent-passport sandbox my agent?

No. It's the identity/attestation layer, not a sandbox. It doesn't intercept calls or restrict the process. You enforce assertScope at your tool boundary and log signed actions to your audit store. See patterns.

How is this different from an API key?

An API key proves a caller has a secret. A passport proves which operator vouched for which agent, with which scopes, and each action is signed by the agent — so you get a tamper-evident audit trail, not just access.

There's no revocation with did:key — is that a problem?

For long-lived agents, yes. Mitigate with short expiresIn and re-issue, or use a did:web operator plus a status list for revocable, long-lived setups.

Can a compromised app server forge an action?

Not without the agent's private key. verifyAction checks the signature against the agent's DID, so forged or altered actions fail verification.

How do verifiers decide which operators to trust?

Keep an allow-list of trusted operator DIDs and reject passports issued by anyone else. Trust is rooted in which operators you accept.

Does it work offline?

Yes — did:key resolves locally, so issuing and verifying need no network or ledger.