Ledger / Feb 2026
Malicious OpenClaw skills on ClawHub deliver AMOS Stealer
Strong
Koi Security found the poisoned skills; the case reaches us through the AI Incident Database.
- Model
- Not namedOpenClaw
- Agent harness
- OpenClawopen-source harness
- Attributed country
- Not attributedthe source names no country
- Sectors hit
- software supply chain, individualsOpenClaw agent users
Attackers published hundreds of poisoned OpenClaw skills on ClawHub whose instructions got users and their agents to install the AMOS stealer. In scope as an agent hijack: the victim's agent ran the payload.
“Bitdefender researchers reported abuse in OpenClaw's third-party 'skills' ecosystem.”AI Incident Database
Sources
- Malicious OpenClaw skills on ClawHub deliver AMOS StealerReported via AI Incident DatabaseAI Incident Database · Feb 2026
- Date
- Feb 2026
- Actor
- ClawHavoc skill publishers (criminal), as attributed by AI Incident Database
- Targets
- OpenClaw agent users
- What the AI did
- Agent hijack, Supply chain
- Model as named
- OpenClaw
- Model basis
- A researcher names it
- Autonomy
- Assisted a human
- Outcome
- Users and their agents installed the AMOS stealer and other credential-stealing malware.