Detection Scoring
How LobsterHoney's scoring engine classifies visitors as human, bot, AI agent, or malicious AI agent.
LobsterHoney uses a multi-signal scoring engine to classify every session with a threat score, classification, confidence level, and severity rating.
How Scoring Works
Every time a visitor interacts with a trap, the scoring engine evaluates the accumulated signals for that session. Signals are divided into two categories: tripwire signals (high-confidence indicators that only AI agents would trigger) and behavioral signals (patterns consistent with automated activity).
Each signal contributes to the session's total score. The score determines the classification.
Signal Categories
Tripwire Signals
These are definitive indicators. A human visitor would almost never trigger these:
- Callback hit -- The visitor followed a hidden callback URL embedded in trap content
- Extraction success -- The visitor's instructions or identity were captured
- Credential use -- A canary credential was used to authenticate against a monitored endpoint
A single tripwire signal is strong evidence of non-human activity. Multiple tripwire signals provide near-certain confirmation.
Behavioral Signals
These are patterns that suggest automated activity but are not individually conclusive:
- Credential extraction -- The visitor extracted credential-like strings from trap content
- Crawl patterns -- The visitor accessed multiple traps in patterns consistent with automated scanning
- Request timing -- Timing and ordering of requests suggests automation rather than human browsing
Behavioral signals alone produce a BOT classification. Combined with tripwire signals, they strengthen the overall confidence level.
Classification Levels
Each session is assigned one of four classifications based on its accumulated signals:
| Classification | Meaning |
|---|---|
HUMAN | Likely a human visitor. Minimal suspicious signals. |
BOT | Likely a traditional bot or crawler. Some automated patterns detected. |
AI_AGENT | Strong evidence of an AI agent. Tripwire signals present. |
AI_AGENT_MALICIOUS | Confirmed malicious AI agent. Multiple high-confidence signals including credential extraction or use. |
Classification thresholds are tuned to minimize false positives. A session classified as AI_AGENT_MALICIOUS has triggered multiple high-confidence tripwire signals -- this is a strong indicator, not a guess.
Confidence and Severity
Each session also receives:
- A confidence score (0--100%) reflecting how much evidence supports the classification. Higher confidence means more signals from more diverse categories.
- A severity rating (Low, Medium, High, Critical) reflecting the potential impact of the detected activity. Sessions involving credential use or intelligence extraction receive higher severity.
Scoring in Practice
The scoring engine runs automatically after every trap hit. As more signals accumulate for a session, the score and classification may escalate. A session that starts as BOT can be reclassified to AI_AGENT_MALICIOUS as the agent continues to interact with traps.
You can view the full scoring breakdown for any session in the Incidents view of the dashboard, including which signals fired and the resulting classification.