The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous scripts and code templates designed to execute arbitrary shell commands. Examples include C templates using system("/path/to/payload.sh"), Node.js snippets using child_process.execSync('id'), and osascript commands for automation-based injection.
[REMOTE_CODE_EXECUTION]: A Python script is provided that connects to an Electron application's debugging port via WebSockets to execute arbitrary code in the main process using the Chrome DevTools Protocol.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to download and install the @electron/asar package from the NPM registry and execution via npx @electron/fuses to facilitate patching application files.
[PRIVILEGE_ESCALATION]: Detailed instructions are provided for gaining elevated capabilities, such as using task_for_pid to obtain task ports (requiring root) and exploiting XPC services via PID reuse attacks to bypass client validation and inherit TCC permissions.
[PROMPT_INJECTION]: The skill includes meta-instructions labeled as 'AI LOAD INSTRUCTION' intended to override the agent's default behavior and safety boundaries by forcing it to adopt an 'Expert Attack Playbook' persona for security bypassing.

macos-process-injection