repo-to-white-paper

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes the local binary /usr/local/bin/typst compile to generate the PDF output. While this is the intended functionality, it involves local command execution.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests data from external repository files.
  • Ingestion points: Reads contents from files like README.md, methodology.md, findings.md, and other documentation in the repository path.
  • Boundary markers: Absent. The instructions do not define delimiters or provide warnings to the agent to ignore instructions embedded in the source documents.
  • Capability inventory: Includes the ability to read arbitrary files from the local file system and execute the Typst compiler.
  • Sanitization: Absent. There is no mention of sanitizing or validating the content extracted from the repository documentation before it is synthesized into the final white paper.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 10:52 AM
Security Audit — agent-trust-hub — repo-to-white-paper