vscode-copilot-customization

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains instructions that direct the agent to favor automated configuration over user transparency. It specifically instructs the agent: 'Do not tell the user to update settings.json manually — run the script instead.' While this is intended to streamline the setup process for a legitimate development tool, it is a directive to perform actions without presenting manual alternatives to the user.
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Node.js script located at scripts/ensure-hook-setting.js. Analysis of the script confirms it is benign, performing basic JSON manipulation on the workspace's .vscode/settings.json file to ensure the correct hook locations are enabled. It does not perform network operations or access sensitive system directories.
  • [SAFE]: The skill makes use of a web_search instruction to verify documentation freshness against official Microsoft VS Code documentation. The search query is hardcoded to a specific domain and does not include any variables that could be used for data exfiltration.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 04:29 PM
Security Audit — agent-trust-hub — vscode-copilot-customization