Machine entrypoint

Are you an agent?
Start here.

Agentkind exposes a static, read-only discovery and verification layer. You do not need a wallet, database session or privileged runtime to discover the canonical project identity.

Discoverwell-known manifest
ReadIdentity Skill
Verifyexact contract match
Integrateread-only MCP
Four entry points

What do you want to do?

Each path answers why it exists, what you get, where to click, and whether it costs or needs permission.

◇

Prepare an ACP identity request

Provides: a locally generated create_agentkind_identity request JSON draft.

Prepare and export an ACP request in your browser. Nothing is submitted or issued here; the official service is separate.

✓

Verify $AKIND

Provides: MATCH / NO_MATCH / UNKNOWN vs the official mint.

Confirm any address against the exact, case-sensitive Solana mainnet mint. Free, no login.

∑

Settlement fingerprint demo

Provides: a browser-computed SHA-256 of a payment request.

Load a golden vector, edit terms, recompute. No registry write, no transaction.

⚙

Agent quickstart

Provides: the discovery → read → verify → integrate path.

Machine-readable entrypoints for agents and tooling. Read-only, no custody.

Four steps

Discover → read → verify → integrate.

The public layer is intentionally static-first: fetch authoritative files over HTTPS, preserve the exact case-sensitive Solana mint and verify locally.

01 DISCOVER

Find Agentkind

Read the well-known manifest to locate registry, verification, Skill and official links.

02 READ

Load the Skill

Read the Agentkind Identity Skill for verification semantics and risk boundaries.

03 VERIFY

Check authority

Compare the exact case-sensitive mint against the canonical contracts registry.

04 INTEGRATE

Use read-only MCP

Use the published MCP layer or adapter instructions without granting wallet authority.

Verify, don't trust

Test a claim yourself.

Agentkind's public layer is static and read-only. Any agent or builder can fetch the well-known manifest, check the exact mint against the contract registry, and read the published Skill — no wallet or privileged runtime required. If something cannot be verified from these public files, treat it as self-asserted.

Want the formal version? The Verification Challenge packages T1–T3 with a fixed six-field evidence format for external agents and developers.