AI
Atlas AI
JK
← Template library
OWASP LLM Top 10LLM10:2025mediumv1.0.0 · System

Unbounded Consumption Guard

Throttles requests per user and per application to prevent denial-of-wallet attacks and runaway costs.

📘Clone & start observing

Creates a Guideline policy. Observation only — nothing is blocked until you promote to Strict.

Mode on clone: log
Defaults to template name. Customise to distinguish multiple instances of the same template.
Leave empty to apply broadly via the template's default data-classification / risk-tier filters.
Rationale

LLM tokens cost money. Without rate limiting, a single misbehaving user or compromised account can burn through monthly budgets in hours.

Example violation
Single user sending 500 requests per minute against the Customer Chat Bot
Triggers (1)
  • inputCount requests against rolling window
Detectors (2)
  • rate_counterrequest-rate
    Sliding-window request counter
  • rate_countertoken-rate
    Token consumption per hour
Actions (3)
  • blockReject with HTTP 429
  • logRecord throttle event
  • notifyAlert if sustained throttling indicates attack
Tunable parameters (3)
Max requests per user per minute
basicnumber
30 is generous for human users; tune lower for kiosks.
Default: 30
Max tokens per user per hour
basicnumber
Adjust based on typical use case complexity.
Default: 100000
Notification channel
basicchannel
Where to send throttle alerts.
Default: "#ops-alerts"
Template defaults (suggested target after promotion)
Suggested mode
block
Risk tiers
Data classifications
Departments

Cloned policies start in Guideline mode. Use the promotion wizard to flip to Strict once you trust the false-positive rate.