lov-hanzi-lens
Warn
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill is designed to interact with a mandatory dependency (lov-professional-infographic) by locating and executing its CLI script (infographic_cli.py) via the subprocess module. The path to this script is resolved at runtime based on environment variables or installation directories. \n- [COMMAND_EXECUTION]: The main script scripts/hanzi_lens.py executes system utilities such as fc-list and the external dependency's CLI to perform font discovery and infographic rendering. The skill also reads user profile configuration files from the home directory to resolve brand assets. \n- [EXTERNAL_DOWNLOADS]: Installation instructions require cloning external repositories from example.com and downloading browser binaries via playwright install chromium. \n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests character data and definitions from external web sources (e.g., zdic.net, ctext.org). While the skill implements safety filters to prevent the generation of content related to personal fortune-telling or naming analysis, the interpolation of untrusted data into instructions remains a potential risk. \n
- Ingestion points: Dictionary data retrieved from external URLs and user-supplied character inputs in scripts/hanzi_lens.py. \n
- Boundary markers: The skill instructions in SKILL.md include explicit directives to distinguish historical fact from modern interpretation. \n
- Capability inventory: The skill can execute shell commands, write files to the project directory, and render HTML to images. \n
- Sanitization: Employs Unicode normalization and regex-based filtering to remove placeholder text and restricted semantic topics.
Audit Metadata