applicationinsights-web-ts
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [Indirect Prompt Injection Surface]: The skill implements telemetry collection that captures untrusted data from the browser, including user interactions, URL parameters, and GenAI conversation details. While this is necessary for its intended purpose of monitoring, it creates a surface where malicious input could be recorded into telemetry logs.
- Ingestion points: Browser-side telemetry captures user interactions (clicks), custom events, and GenAI messages, tool arguments, and results (
SKILL.md,references/agent-traces.md). - Boundary markers: The documentation does not specify explicit boundary markers for the data recorded, but notes that GenAI content capture is an opt-in feature.
- Capability inventory: The skill uses standard web APIs (fetch/XHR) to transmit data and involves the installation of official SDK packages via npm (
SKILL.md). - Sanitization: The skill provides examples of using telemetry initializers to scrub PII and secrets from query strings using regex patterns and encourages gating sensitive content capture behind runtime flags (
SKILL.md,references/agent-traces.md). - [External Resource Downloads]: The skill facilitates the installation of the core telemetry SDK and various framework-specific plugins from the official package registry.
- Evidence: Instructions include
npm i --save @microsoft/applicationinsights-weband related plugins for React, Angular, and React Native. - Context: These resources are standard dependencies for the instrumentation functionality described and originate from a well-known service provider.
Audit Metadata