news-aggregator-skill-2
Warn
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to dynamically construct shell commands (e.g.,
python3 scripts/fetch_news.py --keyword "...") using user-supplied keywords. This creates a surface for command injection if keywords contain shell metacharacters.\n- [PROMPT_INJECTION]: The skill includes a trigger phrase ('news-aggregator-skill 如意如意') that instructs the agent to read and display specific files, potentially bypassing intended constraints.\n- [PROMPT_INJECTION]: Indirect Prompt Injection Risk:\n - Ingestion points: External article content retrieved through the
--deepflag infetch_news.py.\n - Boundary markers: Absent; the instructions lack delimiters or directives to ignore instructions within the fetched content.\n
- Capability inventory: Ability to execute shell commands via
python3and write files to thereports/directory.\n - Sanitization: Absent; the retrieved content is processed for analysis without prior validation or filtering.
Audit Metadata