ProofClaw

Proof that agents can be trusted.

AgentShield notarizes your agent, issues a ProofCard, and enforces runtime policy — so platforms and users can verify trust in one click.

What is ProofClaw?

Signed provenance

Every agent gets a cryptographically signed trust card that binds identity, version, and permissions to a verifiable record.

Runtime enforcement

Shield policies gate network access, secret handling, and tool risk — enforced at deploy time, not just review time.

Verifiable badges

Embed a trust badge in your README, marketplace listing, or chat UI. Anyone can verify it against the transparency log.

How it works

  1. 1

    Publish

    Register your agent and declare its capabilities.

  2. 2

    Notarize

    An independent notary reviews and signs a ProofCard.

  3. 3

    Enforce

    Runtime policies gate what the agent can actually do.

  4. 4

    Verify

    Anyone can verify the badge against the transparency log.

Trust primitives

ProofCards

Portable JSON attestations that travel with your agent. Machine-readable, human-auditable.

Badge verification

Lightweight SVG badges backed by Ed25519 signatures. Verify in the browser or on the command line.

Transparency log

Append-only, hash-chained record of every notarization, revocation, and key rotation. Coming soon: public read access.

Launch with proof.

Get your agent verified, earn a ProofCard, and ship with a badge your users can trust.