pixiv
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill retrieves and displays untrusted metadata from Pixiv, such as illustration titles and descriptions. This content, created by third parties, represents a surface for indirect prompt injection.
- Ingestion points: External metadata is ingested via the search and ranking results in scripts/pixiv-cli.js.
- Boundary markers: No specific delimiters or warnings are used to isolate untrusted content from the agent's instructions.
- Capability inventory: The agent can execute commands (to run the CLI) and perform file system operations (via the download functionality in scripts/pixiv-cli.js).
- Sanitization: The skill outputs raw metadata fields without filtering or sanitizing strings for embedded instructions.
- [EXTERNAL_DOWNLOADS]: The skill fetches image content and zip archives from official Pixiv domains (www.pixiv.net and app-api.pixiv.net) to fulfill user-initiated download requests.
- [COMMAND_EXECUTION]: The skill utilizes several Node.js scripts to handle API communication and media processing, requiring the agent to execute these scripts in the local environment.
Audit Metadata