typst-writer

Pass

Audited by Gen Agent Trust Hub on Sep 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to search for packages and documentation using external services like GitHub and SearXNG, which introduces a surface for indirect prompt injection through search results.
  • Ingestion points: External repository metadata from GitHub and search results from SearXNG.
  • Boundary markers: None defined for the search output processing.
  • Capability inventory: Command execution (gh, typst) and file-system write access for document creation.
  • Sanitization: No instructions provided to sanitize or validate external search content.
  • [DYNAMIC_EXECUTION]: The skill documents the use of specialized Typst packages, specifically pyrunner and jlyfish, which allow for the embedding and execution of Python and Julia code within a Typst document during compilation.
  • [COMMAND_EXECUTION]: The skill provides patterns for using the gh (GitHub CLI) for searching and the typst CLI for document compilation and querying. These tools are used for their intended purpose in a developer workflow.
  • [EXTERNAL_DOWNLOADS]: The skill documents the Typst package system which fetches resources from the official Typst Universe registry on GitHub. These references to well-known services are part of the standard tool functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 29, 2026, 09:03 PM