Prompt Injection Detector
Detects prompt injection attacks in text with confidence scores and security recommendations.
Trust signal
1 of 4 pillars
coverage 25%
Unrated — not enough evidence yet to assign a tier.
Pillars
Signals
Evidence tiers, not an endorsement · trust model v0.1.0 (provisional) · CDP-lane settlement only (D3) · Trust API coming soon
On-chain traction
No on-chain settlement recorded. Unverified = not settled or not yet indexed — never treated as zero.
ERC-8004 agent registry
Not found in the ERC-8004 agent registry. This is the default state — absence is not a negative signal.
Identity & classification
Settlement volume
Per-service timeline not available yet. No settlement recorded for this entity in the indexed period.
Endpoints(1)
Scans text for prompt injection patterns including instruction overrides, jailbreak attempts, role-play manipulations, system prompt leakage attempts, hidden instructions, encoding obfuscation, context manipulation, and privilege escalation. Returns detection result with confidence score and recommendation.