grant-proposal
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill uses a Bash script to search for and execute a local Python utility (
extract_paper_style.py). The path to this utility is computed at runtime by searching the project structure and environment variables likeARIS_REPO. \n- [INDIRECT_PROMPT_INJECTION]: The skill processes research data and literature from external sources viaWebSearchandWebFetchto inform the proposal drafting. Maliciously crafted content in these external sources could theoretically influence the agent's output. \n - Ingestion points: Research summaries, literature survey results, and funded project details retrieved from external grant databases and academic search engines. \n
- Boundary markers: The skill lacks explicit delimiters or instructions to treat external research content as potentially untrusted data. \n
- Capability inventory: The skill is permitted to use
Bash(*),Write, and external LLM processing viamcp__codex__codex. \n - Sanitization: No specific sanitization or filtering logic is described for the external content before it is processed by the agent.
Audit Metadata