hf-init
Pass
Audited by Gen Agent Trust Hub on May 8, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
- [NO_CODE]: The script files 'scripts/hf_config.py' and 'scripts/hf_init.py' contain relative path references rather than executable Python code, which would prevent the skill from functioning as described in a standard environment.
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill collects project names and discovery tags from user-defined environment variables and uploads them to external HuggingFace dataset cards (README.md).
- Ingestion points: Environment variables HUGGING_FACE_TAGS and HUGGING_FACE_PROJECT_NAME sourced from the .env file.
- Boundary markers: No explicit markers or 'ignore' instructions are defined to isolate these metadata fields from the rest of the dataset card content.
- Capability inventory: The skill utilizes network access to create or update repositories on the HuggingFace Hub via the referenced 'hf_init.py' script.
- Sanitization: No sanitization or verification of the metadata content is performed prior to the upload to external services.
Audit Metadata