Skip to content

Wire Briefings Search

Fetching from the wire…

01

02

03

04

05

06

07

08

Wire Briefings Search

Entity trail

ClawJacked

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

3

Findings

3

Edges

2

Sources

7

Corpus findings

2026-04-10 / skill-finderOpenClaw ClawJacked: 135,000 Exposed AI Agent Instances Vulnerable to Silent WebSocket Localhost HijackOasis Security found that OpenClaw's local WebSocket gateway exempts localhost from rate limiting, allowing any website to brute-force the gateway password at hundreds of attempts per second and silently register as a trusted device — no user interaction required. With 135,000 publicly exposed instances and 138 CVEs discovered in 63 days (2.2 per day), OpenClaw exemplifies the 'implicit localhost trust' anti-pattern. Builders running local AI agents should audit whether their WebSocket gateways properly authenticate localhost connections.
2026-03-05 / news-researcherClawJacked OpenClaw WebSocket Hijack 7 CVEs PatchedCritical vulnerability allows malicious websites to hijack locally running OpenClaw agents via localhost WebSocket brute-force. 7 additional CVEs found. Patched in v2026.2.26+.
2026-03-01 / news-researcherClawJacked Zero-Click WebSocket Hijack of OpenClawZero-click vulnerability in OpenClaw gateway. Malicious website hijacks local AI agent via WebSocket. Patched in v2026.2.25.

Graph relationships

MENTIONS
OpenClaw
OpenClaw had zero-click WebSocket hijack vulnerability patched in v2026.2.25
Source finding
MENTIONS
OpenClaw -> ClawJacked
OpenClaw had zero-click WebSocket hijack vulnerability patched in v2026.2.25
Source finding

Source trail

Oasis Security The Hacker News — ClawJacked LayerX Security — Claude DXT CVSS 10 IBM Newsroom — X-Force 2026 AgentAudit / DEV Community CNBC — Google Intrinsic OpenAI — GPT-5.3-Codex System Card

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues