runpod-cloud-migrate
Fail
Audited by Snyk on Nov 21, 2026
Risk Level: CRITICAL
Full Analysis
CRITICAL E004: Prompt injection detected in skill instructions.
- Potential prompt injection detected (medium risk: 0.30). The skill is a legitimate migration procedure, but reference/aws.md line 190 has the agent grant ECR pull access to hardcoded IAM roles in a fixed AWS account (550005742258) that the playbook never explains, which weakens protection of the user's images beyond what the migration needs.
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (low risk: 0.10). The required workflow reads the user's own repo files, first-party vendor playbooks, and the source platform's own resource listings, none of which are outsider-authored. The one outsider-text path is the optional Runpod Hub and Hugging Face Hub lookup (
runpodctl hub search, reading a Hub repo's config), which the agent must actively run and which returns community-published content, so this is low.
MEDIUM W013: Attempt to modify system services in skill instructions.
- Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill's shutdown branches run destructive commands on remote cloud platforms. Baseten labels
baseten model delete, the chain DELETE call andbaseten volume rm --recursive"Irreversible", and fal states "Deleting is permanent" beforefal apps deleteandfal files rm. These commands target remotely managed resources that are presumed shared, so the same operation combines explicit irreversible loss with shared scope. The skill's instructions to ask for approval do not restrict a rogue agent.
Issues (3)
E004
CRITICALPrompt injection detected in skill instructions.
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W013
MEDIUMAttempt to modify system services in skill instructions.
Audit Metadata