ljg-xray-paper
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes the system
datecommand (e.g.,date +%Y%m%dT%H%M%S) to generate timestamps used for report filenames and internal metadata. - [DATA_EXPOSURE]: The skill reads from local files
~/Documents/know/soul.mdand~/Documents/know/memory.mdto access personal worldviews and knowledge. It also writes generated reports to the~/Documents/notes/directory. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it fetches and processes content from external URLs (Arxiv, PDFs, etc.) without sanitization.
- Ingestion points: External research paper URLs, Arxiv HTML pages, and local PDF paths.
- Boundary markers: None identified; the skill directly processes fetched content.
- Capability inventory: File system read/write access and shell command execution (
date). - Sanitization: No explicit sanitization or instruction-filtering is applied to the external content before it is used for analysis.
Audit Metadata