cursor-extension-integration

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for processing external text files (e.g., cursor-extensions.txt) to automate the installation of multiple extensions. This represents a potential surface for indirect injection if the input file is sourced from an untrusted origin.
  • Ingestion points: The cursor-extensions.txt file referenced in the extension sync section of SKILL.md serves as an external data source.
  • Boundary markers: No specific delimiters or warnings are provided to the agent to treat the file content as untrusted or to ignore embedded instructions.
  • Capability inventory: The skill is granted Bash(cmd:*), Read, Write, and Edit tools, allowing it to execute commands based on the file content.
  • Sanitization: The provided shell script snippet while read ext; do cursor --install-extension "$ext"; done does not include validation or sanitization of the extension identifiers before execution.
  • [COMMAND_EXECUTION]: The skill makes extensive use of the Bash(cmd:*) tool to manage the IDE environment.
  • It provides commands for installing, listing, and uninstalling extensions via the cursor CLI (e.g., cursor --install-extension dbaeumer.vscode-eslint).
  • The troubleshooting documentation in references/troubleshooting.md includes a command to remove local extension data using rm -rf ~/.cursor/extensions/[extension-name]/. These are standard maintenance operations for the target development environment.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the download and installation of extensions from external sources.
  • It directs users and the agent to the Open VSX Registry (open-vsx.org) and the Microsoft VS Code Marketplace (marketplace.visualstudio.com). Both are well-known and established services for development extensions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 06:48 AM