axiom-xcode-mcp

Warn

Audited by Gen Agent Trust Hub on Oct 11, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions for installing the axe tool via Homebrew from a third-party GitHub repository (cameroncooke/axe) that is not within the trusted vendor scope.
  • [PRIVILEGE_ESCALATION]: Documentation includes commands for enabling the Xcode headless server using sudo (sudo xcrun mcp-server enable), which requires administrative privileges.
  • [DYNAMIC_EXECUTION]: The skill references and documents the RunCodeSnippet tool, which allows for the runtime compilation and execution of Swift code snippets within the project context.
  • [COMMAND_EXECUTION]: Extensive documentation of command-line operations using xcrun, simctl, and the axe CLI for device and IDE control.
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality.
  • Ingestion points: Reads project files via XcodeRead and processes compiler/build logs via GetBuildLog.
  • Boundary markers: No explicit instructions or delimiters for isolating external data from instructions are mentioned in the documentation.
  • Capability inventory: The skill facilitates file modification (XcodeWrite), code execution (RunCodeSnippet), and UI automation (axe).
  • Sanitization: There is no mention of sanitization or validation of the content read from external project files before it is processed by the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 11, 2026, 05:55 AM