who-blue-books
Pass
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
who-api-clientCLI tool via theuvtool manager. This tool is the primary mechanism for accessing the WHO Classification of Tumours API. - [COMMAND_EXECUTION]: The skill workflows involve executing multiple CLI commands using
who-api-clientto search, retrieve content, and download images. These commands are dynamically constructed based on user input and retrieved chapter data. - [PROMPT_INJECTION]: The skill processes content from the WHO API, which introduces a potential surface for indirect prompt injection.
- Ingestion points: Pathology text, chapter headings, tables, and figure legends retrieved from the external WHO database.
- Boundary markers: The agent is instructed to read the content directly without the use of delimiters or 'ignore' instructions to prevent the execution of embedded prompts.
- Capability inventory: The agent can execute shell commands via the
who-api-clientutility. - Sanitization: There is no evidence of sanitization or filtering of the retrieved medical content before it is processed by the agent.
Audit Metadata