claude-code-wrapper
Warn
Audited by Gen Agent Trust Hub on Feb 19, 2026
Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- Data Exposure & Exfiltration (HIGH): The script
scripts/claude_code.shaccesses~/.claude/config.jsonto extract and export theANTHROPIC_AUTH_TOKEN. Accessing sensitive credential files in the home directory is a high-risk operation. The severity is mitigated to MEDIUM for the final verdict as this is the primary stated purpose of the skill. - Indirect Prompt Injection (LOW): The skill creates a surface for indirect prompt injection by passing unsanitized user input directly to the
claudeCLI. - Ingestion points: User input is accepted via command-line arguments (
$@) inscripts/claude_code.sh. - Boundary markers: Absent; user input is not delimited from instructions.
- Capability inventory: The script executes the
claudebinary with inherited environment variables. - Sanitization: No escaping or validation is performed on the input arguments before execution.
- Command Execution (LOW): The script uses
exec claude "$@"to launch the Claude Code interface. While this is the intended behavior, executing commands with direct user input without validation is a security best-practice violation.
Audit Metadata