The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a variety of command-line tools for iOS binary analysis and device communication, including otool, class-dump, ideviceinstaller, and libimobiledevice. It also manages frida-server lifecycle via SSH commands (killall -9 frida-server) as part of its dynamic analysis methodology.
[EXTERNAL_DOWNLOADS]: Fetches the frida-ios-dump utility from the AloneMonkey GitHub repository to perform FairPlay decryption of Mach-O binaries. It also incorporates the @anthropic/mobile-mcp Node.js package for automated UI interactions and screen capture on iOS simulators and devices.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted data from decrypted applications, network traffic captures (Burp Suite), and application screenshots.
Ingestion points: Decrypted Mach-O binaries, HTTP/gRPC/WebSocket traffic logs, and screenshots of application UI (via Mobile MCP).
Boundary markers: None identified.
Capability inventory: Execution of shell utilities for binary analysis and Frida/Objection scripts for runtime instrumentation.
Sanitization: None identified. However, this risk is inherent to the skill's primary purpose of security auditing.
[SAFE]: The automated scanner alert for attacker.example.com refers to a test vector located in payloads/url_scheme_tests.txt. This URL is used to test for server-side redirection and deep-link vulnerabilities during pentesting and does not represent an exfiltration or phishing attempt by the skill itself.

ios-pentest