nlweb-retrieval-backends
Pass
Audited by Gen Agent Trust Hub on Jun 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches documentation and configuration files from the nlweb-ai GitHub repository to maintain up-to-date setup guidance and configuration schemas.
- [COMMAND_EXECUTION]: Utilizes Bash to run local Python modules such as data_loading.db_load for database indexing and performs environment diagnostic checks. These are standard administrative actions for the skill's intended purpose.
- [PROMPT_INJECTION]: The skill manages a surface for indirect prompt injection through the processing of retrieved data. 1. Ingestion points: Remote backends (Azure AI Search, Bing Web Search) and external documentation fetched from GitHub. 2. Boundary markers: Absent; no instructions are provided to delimit or sanitize retrieved content within the agent's context. 3. Capability inventory: Access to Bash, Write, Edit, and WebFetch tools across various scripts. 4. Sanitization: Absent; the skill does not specify validation or filtering procedures for data retrieved from external sources before it is processed.
Audit Metadata