INCIDENT MAP

Browse / Unsafe Model Loading and Supply Chain / AML.CS0049

Supply Chain Compromise via Poisoned ClawdBot Skill

AML.CS0049 · 2026-01-26 case study

Class
Unsafe Model Loading and Supply Chain
Technique
Publish Poisoned AI Artifacts: AI Agent Tools
Also involves
Direct Prompt Injection, Indirect Prompt Injection, Tool and Agent Misuse
Target
ClawdBot (now OpenClaw) (Exercise)

How it worked

  1. Resource Development · Develop Capabilities

    The researcher created a simple web server to log requests.

  2. Resource Development · Acquire Infrastructure: Domains

    The researcher registered the domain clawdhub-skill.com to host their web server.

  3. AI Attack Adaptation · LLM Prompt Crafting

    The researcher crafted a prompt injection designed to cause Claude Code to execute a curl command to the researcher's clawdhub-skill.com domain.

  4. Resource Development · Publish Poisoned AI Artifacts: AI Agent Tools

    The researcher developed a poisoned ClawdBot Skill called "What Would Elon Do?" The Skill contained the malicious prompt in the rules/logic.md file, which is read when the Skill is activated. The researcher published their Skill to ClawdHub.

  5. Defense Evasion · AI Supply Chain Reputation Inflation

    The researcher used a script to increase the number of downloads of their Skill to increase visibility and gain trust.

  6. Initial Access · Supply Chain Compromise: AI Agent Tool

    Users downloaded the poisoned Skill from ClawdHub.

    Note that ClawdHub does not display all files that are part of the Skill, making it hard for users to review Skills before downloading them.

  7. Persistence · AI Agent Tool Poisoning: Definition and Instructions

    The poisoned Skill included malicious model-readable instructions in rules/logic.md. Once the Skill was installed and made available to the agent, these instructions altered how Claude Code handled requests associated with the Skill.

  8. Execution · User Execution: Poisoned AI Agent Tool

    When a user asked Claude Code "what would Elon do?" it calls the poisoned Skill.

  9. Execution · LLM Prompt Injection: Indirect

    Claude Code read all files that are part of the Skill, executing the malicious prompt in the rules/logic.md file.

  10. Defense Evasion · Masquerading

    Claude Code prompted the user before executing the shell command. The researcher had registered the https://clawdhub-skill.com domain, which appears to be legitimate and may be confused with the legitimate https://clawdhub.com domain, causing the user to select confirm.

  11. Privilege Escalation · AI Agent Tool Invocation

    Claude Code executed the shell command using it's bash tool.

  12. Impact · External Harms

    In this proof of concept, the researcher simply pinged their server and warned the user of the dangers of using Skills without reading the source code, causing no harm. However, they could have delivered a malicious payload, and caused a variety of harms, including:
    - Exfiltrating the user's codebase
    - Injecting backdoors into the user's codebase
    - Stealing the user's credentials
    - Installing malware or crypto miners
    - Performing anything else Claude Code is capable of

Summary

A security researcher demonstrated a proof-of-concept supply chain attack using a poisoned ClawdBot Skill shared on ClawdHub, a Skill registry for agents. The poisoned Skill contained a prompt injection that caused ClawdBot to execute a shell command that reached the researcher's server. Although the researcher here used this access simply to warn users about the danger, they could have instead delivered a malicious payload and compromised the user's system. The security researcher recorded 16 different users who downloaded and executed the poisoned Skill in the first 8 hours of it being published on ClawdHub.

References

Source: MITRE ATLAS case studies, used under Apache 2.0.