---
title: "Malicious downloads can trick AI coding agents into compromising themselves"
type: signal
url: https://frontiernewswire.com/s/malicious-downloads-can-trick-ai-coding-agents-into-compromising-784c0d5e
published: 2026-10-26T10:35:08.000Z
byline: "Source: CSO Online"
---

# Malicious downloads can trick AI coding agents into compromising themselves

Source: CSO Online · Published 2026-10-26 · Canonical: https://frontiernewswire.com/s/malicious-downloads-can-trick-ai-coding-agents-into-compromising-784c0d5e

Malicious instruction files planted in a computer’s Downloads folder can trick AI coding agents into executing harmful code or exposing authentication tokens, Forever Security said after blocking an attempted attack on a customer’s machine and demonstrating similar techniques against OpenAI’s Codex and Anthropic’s Claude Code. In its research report, the firm said it discovered a malicious AGENTS.md file in a customer’s Downloads folder that instructed Codex to install a malicious npm package called function-color. The installation was blocked before the malware could execute. The researchers could not determine how the original file reached the computer, but demonstrated that a malicious website could trigger a similar download without a separate approval prompt. The attack exploits the way AI coding assistants process project-specific instructions. Codex uses AGENTS.md, while Claude Code supports CLAUDE.md. Both can interpret instructions contained in these Markdown files when operat

## Sources

- [Read the original at CSO Online](https://www.csoonline.com/article/4240222/malicious-downloads-can-trick-ai-coding-agents-into-compromising-themselves.html)

## Related

- [Modern AI panic reflects "cognitive mercantilism", a zero-sum worldview where intelligence is hoarded as a scarce supply-side commodity managed by elites (Venkatesh Rao/Contraptions)](https://frontiernewswire.com/s/modern-ai-panic-reflects-cognitive-mercantilism-a-zero-sum-worldview-3823c3ba)
- [A look at OpenAI's and Anthropic's aggressive price cuts between June and October, driven by low switching costs and heavy competition from open-weight models (Max Woolf's Blog)](https://frontiernewswire.com/s/a-look-at-openais-and-anthropics-aggressive-price-cuts-between-june-ec34fa67)
- [Meet the AI Agents Trying to Be Real-Life Journalists](https://frontiernewswire.com/s/meet-the-ai-agents-trying-to-be-real-life-journalists-cd0e6f60)
- [What EvilTokens teaches us about defending against AI-assisted attacks](https://frontiernewswire.com/s/what-eviltokens-teaches-us-about-defending-against-ai-assisted-attacks-906f07cb)
- [Elvis Presley's rightsholders sue Anthropic over 274 songs allegedly used to train its models, seeking $41M+; Anthropic says it disagrees and will defend itself (Derrick Bryson Taylor/New York Times)](https://frontiernewswire.com/s/elvis-presleys-rightsholders-sue-anthropic-over-274-songs-allegedly-978a058c)

---

[More Latest Signal](https://frontiernewswire.com/) · [Signals](https://frontiernewswire.com/verified-reporting)