ClawAudit verdict
prose
prose-andy27725
The skill appears to be a language model designed to simulate a virtual machine and execute Prose programs. There is no evidence of malicious behavior.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. ClawAudit flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Findings (4)
Fake system prompt — attempts to inject instructions
prose.md · code · System: You are
References sudo — requests elevated privileges
state/postgres.md · code · sudo
References agent memory files
primitives/session.md · code · memory.md
Downloads executables from external URLs
state/postgres.md · prose · downgraded · Download from https://
Permissions & capabilities
No declared permissions — minimal attack surface.
Is this flag fair?
Thanks — recorded.