claude-cli

Fail

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs users to download installation and update scripts from official Anthropic domains (claude.ai and code.claude.com).
  • [REMOTE_CODE_EXECUTION]: Instructions include piping remote scripts directly to bash or PowerShell interpreters for automated setup of the CLI tool.
  • [COMMAND_EXECUTION]: Multiple examples demonstrate local shell execution for authentication, setting permissions, and running the Claude tool on project files.
  • [PROMPT_INJECTION]: The skill facilitates processing untrusted repository data and command outputs, creating a surface for indirect prompt injection.
  • Ingestion points: External data enters through piped input or filesystem reads when running the claude command.
  • Boundary markers: No delimiters or warnings are used to isolate untrusted content from the agent's instructions.
  • Capability inventory: The claude binary possesses broad permissions including filesystem access and subprocess execution.
  • Sanitization: No data sanitization or validation steps are provided for external content.
Recommendations
  • HIGH: Downloads and executes remote code from: https://claude.ai/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Jul 8, 2026, 04:18 PM
Security Audit — agent-trust-hub — claude-cli