AgentAvow™
BrowseGet AgentAvowFor teamsDocs
Sign inCheck a tool
Research

What we're learning about agent-tool safety.

Quarterly reports and open specifications — every number backed by signed, reproducible scan evidence you can recompute yourself. We publish the methodology, not just the conclusions.

Reports

LiveQ2 2026

State of Agent Security 2026

Agent distribution surfaces + the signed-evidence substrate. 35k+ tools scanned, reproducible.

Read the report →
LiveAug 2026

State of Agent Security — Q3 2026

Of 17k+ agent tools scanned, 34% carry a high or critical finding. The tool-safety blind spot, sized (live figure).

Read the report →

Open specifications

CTEF — Cryptographic Trust Evidence Format
↗
The signed, reproducible evidence envelope. Versioned in the open.
Conformance test vectors
↗
Validate your own implementation against ours.
Public signing keys (JWKS)
↗
Verify any attestation offline.
AgentAvow™

A trust score you can verify · © 2026 AgentAvow

Available in Claude and ChatGPT

Check

BrowseThe IndexCertifiedSandbox

Build

For tool authorsBadgesList your toolPreflightAPIVerify keys

About

How it worksResearchFAQCommunity ↗Feedback ↗TermsPrivacy