openclaw
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from an external website to guide agent actions, which could allow instructions in the fetched content to influence the agent.
- Ingestion points: Documentation files such as
llms.txtand topic-specific pages (e.g.,cli/,gateway/,install/) fetched fromhttps://docs.openclaw.ai/using theWebFetchtool. - Boundary markers: The instructions lack explicit delimiters or warnings to the agent to treat external content as data and ignore any embedded instructions.
- Capability inventory: The skill description includes high-risk capabilities including
CLI commands,deploymenttasks (Docker, Fly.io, etc.), andexectools, which could be triggered or misconfigured by malicious documentation content. - Sanitization: There is no indication of content sanitization or validation before the agent interprets the retrieved documentation.
- [EXTERNAL_DOWNLOADS]: The skill fetches documentation and configuration guidance from the external domain
docs.openclaw.ai. This introduces a dependency on external third-party infrastructure for its core operations.
Audit Metadata