Prompt extraction guardrail missing
ID |
prompt-extraction-guardrail-missing |
Severity |
high (catalogue Medium; platform severity model has no 'medium') |
Remediation Complexity |
trivial |
Remediation Risk |
low |
Remediation Effort |
medium |
OWASP LLM |
LLM07:2025 — System Prompt Leakage |
OWASP ASI |
ASI01:2026 (secondary ASI09:2026) |
Family |
LLM07 — System Prompt Leakage |
Red-team vectors |
Prompt Injection |
Tags |
ai_security |
Description
A public / external-facing agent has no guardrail carrying the prompt_extraction_detection capability. Nothing detects attempts to extract the system prompt, so an extraction attack against a public surface goes unnoticed.
Scoped to public / external surfaces: on an internal / unknown surface (unknown ≠ public) the missing capability is not a finding, and an agent wired to a prompt-extraction / jailbreak-detection guardrail is not flagged.
Examples
agent = create_react_agent(llm, tools)
app = FastAPI() (1)| 1 | Public-facing agent (FastAPI surface) with no prompt-extraction guardrail — flagged. |
Wiring a prompt-extraction guardrail (PromptGuard, rebuff, a prompt-leak detector) produces no finding; a non-public surface produces no finding.