Observed arrival · 2026-11-12
GAVEL lets you test an agent—and audit the receipt
A no-login sandbox runs AI agents against sample data and shows which actions its policies allow or block.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Teams testing AI agents against operational policies
- Worth noticing
- The page specifies Ed25519 signatures, SHA-256 roots, RFC 8785 canonicalization, and an offline verifier.
Field notes
The interface separates scenario runs from receipt checks, with controls to verify a receipt, verify a whole run, or tamper with a receipt and test it. Its policy examples include mail restricted to an allowed recipient domain and documents limited to .docx files beneath a specified root. The page says production receipts can be checked against GAVEL’s published issuer key; sandbox log records use a separate demo key and hash chain.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue