checkpoint
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: This skill presents an attack surface for indirect prompt injection because it is designed to ingest and process data from external threat intelligence and reputation services.
- Ingestion points: External data from reputation services and threat intelligence feeds (SKILL.md).
- Boundary markers: No specific delimiters or boundary markers are defined for these inputs.
- Capability inventory: Includes access to firewall policies, gateway diagnostics, and GAIA OS commands.
- Sanitization: The instructions do not define sanitization or filtering protocols for external data.
- [COMMAND_EXECUTION]: The skill requires the user to run a local initialization script
./scripts/checkpoint-enable.sh, which is a routine setup operation for the platform's integrations.
Audit Metadata