make-playbook
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill implementation involves reading and writing a
MAKE.mdfile in the project root to maintain stage-specific state. This architecture introduces an indirect prompt injection vector, as the agent may process and follow instructions embedded in this untrusted file. The skill includes ethical guidelines inethos.mdto mitigate some risks, but no formal sanitization is performed on the file content. - Ingestion points: Workflow step 1 in
SKILL.mdreads the project'sMAKE.mdfile. - Boundary markers: Absent.
- Capability inventory: File system read and write access for the
MAKE.mdtracker. - Sanitization: None.
- [EXTERNAL_DOWNLOADS]: The
README.mdfile includes instructions to install a tool usingnpxfrom the NPM package@tamng0905/builder-essential-skills. While this is a standard installation method, it involves downloading and executing code from an external repository managed by an individual user account rather than a well-known organization.
Audit Metadata