Run availability and economic abuse

1. 01

   Concepts & system design

   60 min· read · map the mechanismtimer

   R03 — OWASP API4 Resource Consumption

   R04 — OWASP LLM10 Unbounded Consumption

   Read the assigned sections; map the mechanism, threat model and trade-offs.

2. 02

   Project implementation

   120 min· build · integrate · debugtimer

   Execute cross-project availability, denial-of-wallet, and model-extraction abuse tests: oversized payloads, context stuffing, recursive tool loops, retry storms, worker queue saturation, and systematic extraction probing.

   Build, integrate, debug and improve the increment in the active project.

3. 03

   Security lab

   60 min· attack your own systemtimer

   Verify per-tenant cost caps, circuit breakers, rate limiters, and concurrency governors trip cleanly across P1, P2, and P3.

   Test the abuse case, failure path, authorization bypass or resource limit.

4. 04

   Verification & evidence

   30 min· test · record evidencetimer

   Cross-project availability and economic-abuse test report.

   Add or run regression tests; record the 9-point evidence record.

5. 05

   Deliberate practice & Socratic review

   30 min· explain aloud · update gapstimer

   Answer the 8 Socratic questions without notes: What? Why? How? What if? Why not? How can I prove it? Where can it break? What have I learned?

   Re-explain the hardest concept aloud, inspect one edge case, update the gap log.

6. total5 h

• R03OWASP API Security Top 10↗ owasp.org

  BOLA (API1), Broken Auth (API2), Unrestricted Resource Consumption (API4), SSRF (API7), Misconfiguration (API8).

• R04OWASP Top 10 for LLM Apps↗ genai.owasp.org

  Prompt injection, sensitive info disclosure, improper output handling, excessive agency, vector weaknesses, unbounded consumption.

Read only the named sections needed for today's task. Reading sits inside the 60-minute concepts block; it is not extra time.

• docsOWASP GenAI Security Project
  OWASP LLM10: 2025 Unbounded Consumption

  Defines the denial-of-service and denial-of-wallet failure modes for LLM apps: oversized inputs, unbounded output, recursive tool loops and cost amplification.

  R04unchecked 2026-11-11
• videoRynaut — Architecting Automation · 7:58
  Denial of Wallet: How Attackers Turn Your LLM Bill Into a Weapon

  Explains how per-token billing turns a cheap request flood into direct financial damage, the mechanism behind your context-stuffing and retry-storm tests.

  R04checked 2026-11-11
• videoFuzzy Labs · 4:15
  Large Language Model Security: Model Extraction Attacks Explained

  Explains how systematic querying reconstructs a model's behaviour and why unlimited cheap inference enables it, which shapes your extraction-probing test design.

  R04R05checked 2026-11-11
• articleOWASP
  API4:2023 - Unrestricted Resource Consumption

  Covers the concrete API-layer limits (payload size, rate, per-request cost, third-party spend) that your worker-queue saturation and oversized-payload tests are checking for.

  R03moved 2026-11-11
• repoGitHub
  NVIDIA/garak

  A scriptable LLM vulnerability scanner whose probes generate large volumes of requests, so you can run sustained campaigns and measure token spend, latency and rate-limit behaviour under load.

  R04checked 2026-11-11

A lab is complete only when it produces evidence. Installing PyRIT, Garak, Presidio or TransformerLens is not completion. Saved on this device only.