1. Verify online, right here
Nothing to install: paste, click verify, see the result on the spot. Perfect for a first try — it's what you just used above.
The JIAOZI Verifier
A free identity checkpoint: hand it an AI agent's identifier and within seconds it tells you who the agent is, whether the credential is genuine, and whether it is still valid right now. No sign-up, nothing to install — use it right below.
Three ways to use it
2. Take it home
Put the verification capability into your own computer or programs and verify whenever you like — that's the code library, command-line tool and query interface below.
3. Put it on your website
Embed the verifier on your site so every visitor can verify agents in place. Each embed is a little signboard that keeps itself up to date.
Four forms — take what fits
Embed code for your website
★★★ Recommended · easiest
Copy a short snippet into your page and a live identity card appears: green while the credential is valid, red the moment it is revoked — and nobody can fake it. Your visitors verify agents without leaving your site.
Get the embed code →Code library (an SDK)
★★★ For developers
Drop a ready-made library into your own system and every visiting agent gets verified automatically — no human in the loop. Available for JavaScript and Python; npm and PyPI listings are in preparation.
Read the integration guide →Command-line tool (a small program you run in a terminal)
★★ For terminal users
What ships today is Validator, a tool that health-checks your own agent and produces a summary of proof information (format code name: jiaozi.attest.v1). Everything runs on your machine; code and data are never uploaded. A command-line verifier for other people's credentials is in the works.
Read the manual →Browser extension & desktop app
☆ Not yet available
Honestly: not built yet. Most agent verification happens automatically inside programs, and people rarely open an app to verify by hand — so these come later. If you genuinely need one, write to us; enough demand puts it on the roadmap.
Write to us →Not a person but a program?
If the one doing the verifying is a program, it can call our public query interface (an API — the channel programs use to talk to each other) directly: put the DID or credential number into the address below and get a machine-readable result. No account needed.
The machine summary for AI lives at /llms.txt; full interface details are in the developer docs. Developer docs →
What it can verify (not just ours)
This verifier is built on the public identity standards set by the W3C (the World Wide Web's standards body). It can run baseline checks on any AI credential issued to those standards — not only ours. For identities we didn't issue, we return a neutral standards-conformance report and endorse nothing.
- It works the other way too: credentials we issue can be verified with anyone's standard tools — even if our website went dark one day, credentials already issued would still verify. We treat that as the litmus test of the word “open”.
- Verification needs only public information: the credential itself and the public key that matches it. Your code and your data never have to be handed to anyone.
Want your agent to be verifiable at a glance, worldwide?
Get it a verifiable identity credential — free during the promotion period.