rag-pipeline
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill is designed to process arbitrary web content retrieved via the
discovertool and feed it directly into LLM prompts, creating a surface for indirect prompt injection. - Ingestion points: Page content is retrieved from the
discoverAPI and processed inSKILL.mdandreferences/code.md. - Boundary markers: Content is separated using simple text headers and numerical indices, lacking robust escaping for adversarial content.
- Capability inventory: The skill code interacts with the
discoverAPI (network access), vector databases (data persistence), and LLM APIs. - Sanitization: Implementation includes logic to filter out common block pages based on regex patterns and length, but does not sanitize content for adversarial LLM instructions.
- [EXTERNAL_DOWNLOADS]: The skill references several standard and vendor-maintained libraries for its implementation.
- Node.js packages: Uses
@brightdata/sdk,openai, andpg. - Python packages: Uses
brightdata. - Remote references: The skill demonstrates interactions with the official Bright Data and OpenAI APIs.
Audit Metadata