Trust Levels

STACK uses a three-layer trust model. Each layer builds on the previous one, enabling graduated trust for different use cases.

L0 -- Base (Any Passport)

The default trust level. Any agent with a valid passport qualifies. No additional verification required.

• Read public data
• Browse skills marketplace
• Create drop-offs between own agents

L1 -- Humanity (Verified Human)

Proves a real human is behind the agent's operator account. Achieved through identity verification providers.

• Access services requiring human verification
• Higher rate limits
• Publish skills on the marketplace
• Required by some skills before invocation

L2 -- Identity (Verified Identity)

Proves the specific identity of the human behind the operator. Document verification + biometric matching.

• Access regulated services (financial, legal)
• Act on behalf of a verified individual
• Enterprise compliance requirements
• Highest trust tier for skill invocation

L3 -- Delegation

A verified human explicitly delegates specific scopes to their agent. The agent carries a delegation claim proving consent.

• Agent acts on behalf of a specific person
• Scopes can only narrow, never widen
• Consent is revocable at any time

Identity Providers

STACK supports multiple identity verification providers:

• BankID Sweden (bankid_se) -- Nordic banking identity, high assurance
• Stripe Identity (stripe_identity) -- Document + selfie verification, high assurance
• Login.gov (login_gov) -- US government identity, IAL1=substantial / IAL2=high
• ID.me (id_me) -- US identity verification, IAL1=substantial / IAL2=high
• Plaid (plaid) -- Financial identity verification, high assurance