5writing

Pass

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes typesetting compilers including xelatex and typst compile via Bash. This is a legitimate and necessary function for converting LaTeX and Typst source files into final PDF documents as part of the paper writing process.
  • [INDIRECT_PROMPT_INJECTION]: The skill acts as a data consumer by reading content from reports/RESULTS_REPORT.md and reports/DRAWIO_REPORT.md to inform the paper's narrative. While this creates an ingestion surface for potentially untrusted data, the risk is negligible within the context of generating academic reports based on a user's own project analysis. • Ingestion points: Local markdown reports residing in the reports/ directory. • Boundary markers: The instructions do not specify explicit delimiters for ingested data, relying on the agent's contextual understanding of the project flow. • Capability inventory: The skill possesses Write and Bash capabilities to produce paper sections and compile the final document. • Sanitization: The skill prioritizes data fidelity from the reports to ensure mathematical accuracy in the generated paper, which is consistent with its primary purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 13, 2026, 01:33 PM
Security Audit — agent-trust-hub — 5writing