Agent Incident RegistryAIR
v1 · updated 19 Oct 2026
Disclosed vulnerability

CVE-2024-12911: prompt injection to SQL injection to arbitrary file creation in LlamaIndex JSONalyzeQueryEngine

AIR-2025-0016 AIR record · 2025-03-llamaindex-jsonalyze-sql-injection
From the supporting source
A vulnerability in the `default_jsonalyzer` function of the `JSONalyzeQueryEngine` in the run-llama/llama_index repository allows for SQL injection via prompt injection. This can lead to arbitrary file creation and Denial-of-Service (DoS) attacks.github.com
The record
Disclosed
2025-03-20
Vendor
LlamaIndex (run-llama)
Product
LlamaIndex (llama-index / llama-index-core)
Risk domain
Tool misuse › Uncontrolled resource consumption
System type
Generative model
Who acted
Agent acted — the engine generated and executed SQL that wrote files
Attack surface
Agent framework
Autonomy level
Semi-autonomous
Initial vector
Prompt injection into generated sql
Impact type
Arbitrary file write, Denial of service
Confidence
High
Submitted by
EAI Internal
Crosswalk
OWASP Agentic Top 10 (2026)
ASI02 – Tool Misuse and Exploitation
MITRE ATLAS
AML.T0029 – Denial of ML Service
NIST AI RMF
MAP 3.2
EU AI Act
—
ISO/IEC
TR 24028; 42001; 23894
AIUC-1
D. Reliability
Scope
tool failure
Mitigation
Impose rate limits, timeouts, and resource quotas on agent actions.
Taxonomy
Enkrypt AI agent risk taxonomy · Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery
Blast radius
applications using JSONalyzeQueryEngine in llama-index < 0.12.3
Capability abused
text_to_sql_generationlocal_sqlite_execution
Defense & remediation
Recorded defense
none - LLM-generated SQL executed directly against a local sqlite instance
Guardrail kinds
None
Guardrail outcome
Not applicable
Fix shipped
llama-index 0.12.3
Taxonomy & references
OWASP ASI01 Agent Goal HijackOWASP ASI02 Tool Misuse and ExploitationCVE-2024-12911prompt_injection_to_sql_injectiontext_to_sqlinsecure_temp_filecwe_379
Curator notes
GHSA-jmgm-gx32-vp4w, CVSS 7.1 High, CWE-379. Reported via huntr. Exploitation runs ATTACH DATABASE to create files on the filesystem. Canonical example of an LLM-generated-query surface in an agent framework.
Cite as
Enkrypt AI. "CVE-2024-12911: prompt injection to SQL injection to arbitrary file creation in LlamaIndex JSONalyzeQueryEngine". AIR, AIR-2025-0016. https://agentincident.vercel.app/i/AIR-2025-0016/ (accessed 2026-10-19).