long-form-article
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts to facilitate its workflow. Specifically, it runs
scripts/scan.pyto perform a mechanical check of the article draft for word count, placeholders, and formatting issues. It also attempts to run a context-gathering script,../start-here/scripts/brand_context.py, which is located in an adjacent directory. These scripts use the standard library and perform benign operations. - [EXTERNAL_DOWNLOADS]: To perform research, the skill leverages external services such as Exa and Firecrawl. These tools are used to fetch studies and verify primary sources. The use of these well-known services for their intended purpose is documented and consistent with the skill's primary function.
- [INDIRECT_PROMPT_INJECTION]: The skill acts as a research assistant, ingesting data from the web (via Firecrawl/Exa) and local files. While the instructions do not explicitly mandate the use of delimiters or specific safety headers to ignore potential instructions embedded in researched content, this is a standard operational surface for research-based agents and is considered a low risk in this context.
Audit Metadata