Caveat verdict
data-boundary
88
🟢 Trusted
No high-risk patterns surfaced by the deep scan — automated capability review, not behavioral proof.
Security boundary skill that parses untrusted CSV/JSON through a tool layer before passing structured results to the model; this is a defensive skill explicitly designed to prevent prompt injection.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. Caveat flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
70
security
90
transparency
70
maintenance
Findings (1)
Pattern match critical
Prompt injection — tries to override agent instructions
references/output-schema.md · code · ignore previous instructions
Permissions & capabilities
No declared permissions — minimal attack surface.
Is this flag fair?
Thanks — recorded.