APort

checked 1 day ago

APort issues W3C DID/VC based passports for AI agents and enforces pre-action authorization policies before tool calls run.

Identity and trustLive

Agents get a passport, a W3C DID/VC based identity document declaring the agent's capabilities, that other systems can verify before authorizing an action, plus cryptographically signed (Ed25519) allow or deny decision records for audit.

Credential type
W3C DID/VC based agent passport
Who accepts it
Claude Code, Cursor, LangChain, LangGraph, CrewAI, OpenAI-based agents, MCP tools
Delegation model
not published
Revocation
not published
Reputation model
Passport tracks declared agent capabilities and a running audit log of signed authorization decisions rather than a public trust score.
Integration
API, SDK, MCP
Pricing
Open source tier is free forever and includes local passports, 21 policy packs, and 7 framework integrations. Pro tier adds hosted passports, a cloud dashboard, and signed decisions, and is free during beta with usage based pricing planned after. Enterprise tier is custom priced and adds self-hosted API, custom policy packs, and SSO.

Our take

Candidate money category was not confirmed: APort enforces spend caps and limits as policy rules on agent actions, but it does not sell a wallet, account, card, or payment rail itself, so money was dropped in favor of identity only.