The Agent Skills Directory

[COMMAND_EXECUTION] (HIGH): Vulnerability in scripts/setup-auth.sh due to unquoted heredoc usage. The shell evaluates variables within << EOF, allowing command substitution if the $API_KEY variable contains backticks or $(). Evidence: Line 79-84 in scripts/setup-auth.sh.\n- [COMMAND_EXECUTION] (HIGH): Vulnerability in scripts/test-connection.sh via unsafe environment loading. The use of export $(...) with xargs on .env file contents allows execution of embedded commands if the file is manipulated. Evidence: Line 17 in scripts/test-connection.sh.\n- [EXTERNAL_DOWNLOADS] (MEDIUM): scripts/install-sdk.sh performs automated installation of dependencies from NPM and PyPI registries (elevenlabs, dotenv) without integrity checks or version pinning.\n- [PROMPT_INJECTION] (HIGH): High risk of Indirect Prompt Injection (Category 8). An attacker can provide a malicious 'API key' through external data that results in local command execution when processed by the agent. 1. Ingestion: scripts/setup-auth.sh via command line arguments. 2. Boundary markers: Absent. 3. Capability inventory: Bash tool, file-write, and network access via curl. 4. Sanitization: Absent; input is directly interpolated into shell commands.

api-authentication