Knowledge Base Ingestion

Warn

Audited by Snyk on Feb 21, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (high risk: 0.90). This skill explicitly fetches and ingests content from user-provided URLs (see the "URL Ingestion" capability and the "Adding Web Content" workflow in SKILL.md), allowing arbitrary public third-party web pages to be read and later influence agent behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

  • Potentially malicious external URL detected (high risk: 0.70). The skill exposes a runtime operation (kb_ingest_url) that fetches arbitrary user-provided URLs and ingests their content into the vector knowledge base, allowing external content fetched at runtime to be injected into the agent's retrieval/context and thus directly influence prompts — flagged: user-provided URLs via kb_ingest_url.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 21, 2026, 03:25 PM