The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill requires the agent to analyze and interpret data from untrusted external files, which could contain adversarial instructions designed to influence the agent's logic.
Ingestion points: The agent processes external data via commands like ipsw macho info, ipsw symbolicate, ipsw dyld disass, and ipsw sb dec across all reference files and the main skill body.
Boundary markers: The instructions do not define specific delimiters or warnings for the agent to ignore potentially malicious strings embedded within binary metadata, symbols, or log files.
Capability inventory: The skill utilizes the ipsw CLI to perform deep file system analysis, header generation (class-dump), and remote firmware extraction.
Sanitization: There is no evidence of sanitization or validation of the content extracted from binaries before it is presented to the agent for interpretation.
[COMMAND_EXECUTION]: The skill is centered around the execution of the ipsw command-line utility to perform reverse engineering tasks.
Numerous examples in SKILL.md and the references/ directory illustrate the execution of shell commands for kernel analysis, dylib extraction, and symbolication.
[EXTERNAL_DOWNLOADS]: The documentation references external resources for tool installation and data acquisition.
The skill suggests installing the ipsw tool via the vendor's Homebrew tap (blacktop/tap/ipsw).
Reference documentation (references/download.md) includes patterns for downloading official firmware components from Apple's content delivery network (updates.cdn-apple.com).
These downloads are associated with legitimate vendor infrastructure or well-known, trusted services.

ipsw