Bearer Token Disclosure detection
A publicly auditable detection pattern in the LeakSnitch engine. Category: Password and Credential Disclosures. Severity: Critical.
Short answer
LeakSnitch detects bearer token disclosure with a base score of 94 (Critical severity). The pattern below is fully public, so any security team can verify exactly what gets flagged before rollout.
(?:here(?:'s| is)|my|the|please use|use (?:this|the))\s+bearer\s*(?:token)?\s*(?:is|:|=|-)?\s*["']?([A-Za-z0-9\-_.=]{20,})Base score
94 / 100
Severity
Critical
Category
Password and Credential Disclosures
Why it matters
This is a natural language disclosure pattern. It fires when someone hands a credential to an AI in plain sentences, which is how most real leaks happen.
In the LeakSnitch engine this pattern is one of 58 public rules combined with context signals, entropy analysis, and checksum validators. A single match never blocks by itself; the full context aware scoring decides whether bearer token disclosure in a prompt is a real leak or a false positive.
Frequently asked questions
Why is the bearer token disclosure pattern public?
LeakSnitch's competitive advantage is the context aware engine, not hidden regexes. Publishing every pattern lets security teams audit, test, and trust the detection before it ships to their organization.
Can I test this bearer token disclosure pattern?
Yes. Paste sample text into the free LeakSnitch scanner in your browser and watch the exact detection in action, including severity and scoring.
Does LeakSnitch use other signals for bearer token disclosure?
This public pattern is the format based layer. The engine adds context signals, disclosure phrasing, entropy fallback, and checksum validation to confirm high confidence before blocking.
Related patterns
Stop AI leaks on the device, not in the cloud.
Install LeakSnitch in under 30 seconds and protect every prompt sent to ChatGPT, Claude, Gemini, and 25+ AI tools. Free for individuals.