The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references external documentation and sample project data from public repositories for testing purposes.
Evidence: Mentions using sample inputs from the rstackjs/rstack-examples repository on GitHub and refers to documentation at reverse.put.as and keith.github.io.
[INDIRECT_PROMPT_INJECTION]: The skill methodology involves processing external sample data to validate sandbox configurations, which introduces a surface for indirect prompt injection if the processed content contains malicious instructions.
Ingestion points: The documentation in SKILL.md suggests fetching example projects from external GitHub repositories during the testing phase.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are defined for the external test data.
Capability inventory: The skill has access to the Bash tool to execute sandbox-exec and the Write tool to create configuration files.
Sanitization: There is no explicit sanitization step mentioned for external content before it is used within the profiling or testing workflow.

seatbelt-sandboxer