Find notable cyber news and cases, enriched with sources, timelines, and signals.
Home Watchlists Beta Browse News Feed Stats About Contact

GuardFall shell-trick bypass of command safety checks in AI coding agents

Technical Analysis
First reported
Last updated
Happening score
H score 25
1 unique sources, 1 articles

Summary

Hide ▲

GuardFall exposed a shell-trick bypass that lets dangerous commands slip past safety checks in open-source AI coding and computer-use agents, putting full account access at risk. The bypass worked against 10 of 11 tested agents and could reach a real shell before the guard understood what would run. Only Continue was built to resist the default attack path.

Related Happenings

Ghostcommit PNG-embedded prompt injection against AI code reviewers

Technical Analysis
H score25 First: 11.07.2026 12:03 Last: 11.07.2026 12:03 Sources 1

About this happening: Researchers demonstrated Ghostcommit, a PNG-embedded prompt-injection technique that can bypass AI code review and leak .env secrets into committed source. The pay...

Open Happening

Prompt-injection proof-of-concept enables silent RCE in Claude Code and Codex

Technical Analysis
H score28 First: 10.07.2026 16:45 Last: 10.07.2026 16:45 Sources 1

About this happening: Researchers demonstrated a proof-of-concept exploit that can force remote code execution in Anthropic’s Claude Code and OpenAI’s Codex, exposing a trust-boundary f...

Open Happening

Enterprise AI deployments need governance and segmentation after red-team failures

Defensive Guidance
H score15 First: 24.04.2026 15:10 Last: 24.04.2026 15:10 Sources 1

About this happening: Enterprise AI deployments are exposing familiar security gaps, making governance, segmentation, and red-team validation urgent to reduce the risk of data theft...

Open Happening

Indirect prompt injection payloads against AI agents reveal fraud, deletion, and secret-theft paths

Technical Analysis
H score20 First: 23.04.2026 12:30 Last: 23.04.2026 12:30 Sources 1

About this happening: 10 new indirect prompt injection (IPI) payloads show how web content poisoning can coerce AI agents into financial fraud, data destruction, and API key theft...

Open Happening

Claude Code deny-rule bypass fix (version 2.1.90)

Security Patch Release
H score17 First: 08.04.2026 12:16 Last: 08.04.2026 12:16 Sources 1

About this happening: Anthropic released Claude Code version 2.1.90 last week to fix a command-parsing flaw that could let user-configured deny rules silently stop applying when a command e...

Open Happening

Timeline

  1. 30.06.2026 17:26 2 articles · 15d ago

    GuardFall shell-trick bypass of command safety checks in AI coding agents

    Initial Disclosure

    A shell-parsing mismatch in AI agent command guards allowed plain-text filters to miss what bash would actually execute. The bypass was demonstrated across most tested open-source coding and computer-use agents and singled out Continue as the only one with built-in resistance.

    Show sources
    Open in new tab