dotnet-playwright

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill facilitates interaction with external web pages, creating a surface for indirect prompt injection. An agent could potentially execute malicious instructions if they are encountered within the content of a navigated page.
  • Ingestion points: Use of Page.GotoAsync() and data extraction via locators in SKILL.md.
  • Boundary markers: No explicit delimiters or instructions are provided in SKILL.md to distinguish untrusted web content from internal agent logic.
  • Capability inventory: Includes full browser automation, file uploads via Page.SetInputFilesAsync(), and network navigation in SKILL.md.
  • Sanitization: No sanitization of page content is described in SKILL.md before processing.
  • [COMMAND_EXECUTION]: Provides instructions in SKILL.md for installing browser binaries using standard pwsh scripts and the dotnet CLI. These operations are part of the official Microsoft Playwright setup process and target trusted tools.
  • [EXTERNAL_DOWNLOADS]: Orchestrates the download of browser binaries (Chromium, Firefox, WebKit) from official Microsoft repositories as described in SKILL.md to enable end-to-end testing functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 03:43 PM