Disclosed vulnerability
CVE-2026-37006: unauthenticated RCE in gpt-researcher WebSocket endpoint via malicious MCP configurations
AIR-2026-0213 AIR record · 2026-08-gpt-researcher-websocket-mcp-rce
From the supporting source
allows an unauthenticated remote attacker to achieve code execution via malicious Model Context Protocol configurations.cveawg.mitre.org
The record
Disclosed
2026-08-27
Vendor
assafelovic / gpt-researcher
Product
gpt-researcher
Risk domain
Access control and permissions › Privilege escalation
System type
Generative model
Who acted
AI was the target — agent framework RCE; no model decision
Attack surface
Agent framework
Autonomy level
Assistant
Initial vector
Unauthenticated network request
Impact type
Remote code execution
Confidence
High
Submitted by
EAI Internal
Crosswalk
- OWASP Agentic Top 10 (2026)
- ASI03 – Identity and Privilege Abuse
- MITRE ATLAS
- AML.T0055 – Unsecured Credentials
- NIST AI RMF
- GOVERN 6.1
- EU AI Act
- Article 13
- ISO/IEC
- TR 24028; 42001; 23894
- AIUC-1
- B. Security
- Scope
- tool misuse
- Mitigation
- Apply least-privilege principle; Regular policy audits.
- Taxonomy
- Enkrypt AI agent risk taxonomy · Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery
Blast radius
gpt-researcher v0.14.7 and earlier
Capability abused
Taxonomy & references
Curator notes
CVSS 9.8 per CVE Brief (no CVSS in the MITRE record). Published 2026-08-27. Autonomous-research agent whose MCP configuration path reaches code execution. PoC/disclosure; no real victim.
Supporting sources
Cite as
Enkrypt AI. "CVE-2026-37006: unauthenticated RCE in gpt-researcher WebSocket endpoint via malicious MCP configurations". AIR, AIR-2026-0213. https://agentincident.vercel.app/i/AIR-2026-0213/ (accessed 2026-10-19).