Observed arrival · 2026-09-26
QED Proof checks the work an AI agent says it did
The tool checks an agent’s claims against the destination—such as a GitHub repository—and issues signed receipts that others can verify.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Teams supervising AI agents that make external changes
- Worth noticing
- A sample failed receipt flags a claimed GitHub commit absent from main; the on-chain anchor is currently on Base Sepolia testnet.
Field notes
Claims can enter through an SDK, an MCP tool, or a Claude Code hook, after which the service checks the destination rather than relying on the agent’s own trace. The page distinguishes failed checks from inconclusive ones: a timeout or revoked permission produces “Couldn't check,” never “Verified.” It says receipts contain fingerprints rather than the content read, and that the verifier is open source; the current log-root anchor is on Base Sepolia testnet.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
$PaidCommerce or pricing visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
ƒJavaScriptBrowser-side code central
One card from the complete issue