dobby-artifacts

Pass

Audited by Gen Agent Trust Hub on Jun 26, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local shell scripts, specifically scripts/check-fast.sh, within the user's repository directories (~/GitHub/adi, ~/GitHub/angie, ~/GitHub/dobby-engine) to validate changes.
  • [PROMPT_INJECTION]: The references/artifact-contract.md file contains a security-aware instruction stating: "Do not put secrets or external credentials in the manifest," which helps prevent accidental credential exposure.
  • [DATA_EXPOSURE]: The skill interacts with local files in the user's home directory (~/GitHub/). This behavior is restricted to the specific task of managing workspace artifacts and does not include any network exfiltration or unauthorized file access.
  • [PROMPT_INJECTION]: Regarding indirect injection surface:
  • Ingestion points: User-provided data used to populate index.html and artifact.json (SKILL.md).
  • Boundary markers: No explicit markers are defined to isolate user-supplied data from the artifact's HTML templates.
  • Capability inventory: Execution of validation shell scripts (references/artifact-contract.md).
  • Sanitization: The skill does not explicitly describe sanitization procedures for user content embedded into the generated HTML/JS artifacts.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 26, 2026, 03:06 PM
Security Audit — agent-trust-hub — dobby-artifacts