Payload scanner

Scan agent output before execution.

Check an untrusted payload against the recipients your application expects. Warden returns a decision, detector reasons, and a transformed payload when sanitization is available.

01 / Untrusted input

Payload and trusted context

Demo scan

Example payload Untrusted payload

Press Ctrl+Enter or Cmd+Enter to run the scan without leaving the editor. 0 / 4,000 characters

Trusted payment recipients

Add known EVM or Solana recipients. Warden compares payment-like addresses in the untrusted payload with this context.

EVM recipients are normalized to lowercase; duplicates are removed.

Curated examples are requested from this Warden instance. You can write a custom payload now.

Your payload is sent to Warden for this scan. It is never placed in a public share URL by this page.

UNKNOWN No scan has run for this payload.

02 / Warden decision

No scan result yet.

Run a scan to see the verdict, detector reasons, and any changes Warden made to the payload.

03 / Caller action

Not run

Review caller action.

Not established

DecisionNot run

Risk severityNot established

Detections0

LatencyNot measured

Response receivedNot run

Sanitized output

Before and after

Untrusted input

Warden output

Analyzer evidence and detector boundary

Boundary: This free route runs implemented fast detectors only. It does not perform the thorough corpus-similarity check, browse linked resources, or prove that an ALLOW payload is safe.

Raw machine-readable response

{}

About this scan

This no-account demo calls POST /api/demo/scan with a 4,000-character payload limit and up to 20 trusted recipients. It runs the implemented fast detectors, not the thorough CORPUS_MATCH check.