The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill processes user-provided $ARGUMENTS to select relevant reference documentation. While it lacks explicit boundary markers, the resulting actions (modifying configuration files) are restricted to the intended purpose of workspace orchestration.
Ingestion points: $ARGUMENTS variable in command/turborepo.md used for task selection.
Boundary markers: Absent.
Capability inventory: Modification of package.json and turbo.json files; execution of turbo run commands.
Sanitization: Absent.
[Unverifiable Dependencies & Remote Code Execution] (SAFE): The documentation mentions several standard ecosystem tools such as syncpack, manypkg, and sherif. These are well-known community tools for monorepo maintenance and are presented as optional best-practice recommendations.
[Data Exposure & Exfiltration] (SAFE): The skill references environment variables like TURBO_TOKEN and GITHUB_TOKEN, but correctly identifies them as secrets to be handled by CI provider vault systems (e.g., GitHub Secrets). No hardcoded credentials or unauthorized network exfiltration patterns were found.

turborepo