# LastID - Identity and Control for AI Agents

> LastID gives every AI agent a verified identity, credentials it can use but never see, rules you set that it enforces, and a tamper-proof record of everything it did. Get started free at https://lastid.co/signup.

## What is LastID?

LastID is cryptographic identity and governance for AI agents. You give an agent a verifiable identity, grant it least-privilege access, hand it credentials it can use without ever seeing them, sign rules it has to follow, and keep a signed record of every action it took. The same identity, rules, and memory follow your agent across tools.

## What you can do with it

- **Give agents API keys without giving them the keys.** Share a credential and the agent mints a single-use, five-minute handle. Your key is injected at the network boundary, so the agent uses it but never sees it. Every use is scoped, rate-limited, and recorded.
- **Set guardrails your agent cannot talk its way past.** Write deny, warn, and rewrite rules, or use signed packs (supply-chain firewall, secrets, production protection). The agent enforces them locally, even offline. You watch which rules fire over time.
- **Prove what your agent did, and which agent did it.** Every agent has a verifiable identity and every action lands in a per-agent, signed, tamper-evident audit chain you can verify.
- **Run least-privilege agents by default.** Each agent carries a bounded credential and refuses anything outside it. Zero standing access to your secrets.
- **Give agents memory that follows them.** Agents remember decisions and preferences across sessions and tools, find them by meaning, and you govern what they keep.

## Runtimes

LastID is portable. Your agent's identity, rules, and memory travel with it across tools, so you are not locked to one runtime.

- **Claude Code** - available now.
- **ChatGPT** - coming soon.
- **Codex** - coming soon.

## How it works

1. **Verifiable agent identity.** Every agent gets its own DID and a verifiable credential (SD-JWT), derived from your master identity and approved once in your LastID wallet. The credential names exactly which resources and actions the agent may use.
2. **Credentials it can use but never see.** The agent discovers shared items, mints a single-use opaque handle, and the local LastID process attaches the real secret at the network boundary. The secret never enters the model's context.
3. **Signed rules it enforces.** You author and sign rules; the agent applies them locally on every action.
4. **A record you can trust.** Every tool call appends to a hardware-key-signed, hash-linked audit chain on your machine. You verify its integrity any time.

## Get the plugin

- Source code: https://github.com/GetTrustedApp/lastid-agent

## Key technical standards

- Decentralized Identifiers (DID)
- Verifiable Credentials and SD-JWT (selective disclosure)
- BIP39 and BIP85 (seed and key derivation)
- Hardware-backed keys (Secure Enclave, StrongBox, TPM)
- Hash-linked, device-key-signed audit chain
- End-to-end encrypted agent-to-operator messaging
- OAuth 2.0, OpenID Connect, OpenID4VCI, OpenID4VP

## Also from LastID: proof of human identity

LastID also proves the human behind a request, stopping impersonation that MFA and caller ID cannot.

- **Stop callback fraud** - confirm the person before a high-value action.
- **Beat deepfakes** - verify the person, not the pixels.
- **Secure the service desk** - prove identity before a reset.
- **Verify communications** - confirm a sender before you connect.

## Pages

- / - Identity and control for AI agents
- /signup - Get started free
- /use-cases/agent-credentials - Give agents API keys without giving them the keys
- /use-cases/agent-guardrails - Guardrails your agent cannot talk its way past
- /use-cases/agent-audit - Prove what your agent did
- /use-cases/agent-least-privilege - Least-privilege agents by default
- /use-cases/agent-memory - Memory that follows your agent
- /use-cases/callback-fraud - Stop callback fraud
- /use-cases/deepfakes - Beat deepfakes
- /use-cases/service-desk - Secure the service desk
- /use-cases/verified-comms - Verify communications
- /architecture - Technical architecture
- /security - Security practices and key storage
- /faq - Answers for builders and security teams
- /about - Origin story and team
- /blog - Writing on agent identity and trust

## Contact

- Website: https://lastid.co
- Source: https://github.com/GetTrustedApp/lastid-agent
- Email: hello@lastid.co