The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external websites via the scripts/audit_site_for_ios.py script. The resulting audit file (SITE_TO_IOS_AUDIT.md) is used to drive the agent's planning and execution logic.
Ingestion points: Data enters the agent's context from external URLs processed by the audit script (referenced in SKILL.md and references/site-to-ios-runbook.md).
Boundary markers: No explicit delimiters or instructions are provided to the agent to disregard instructions potentially embedded in the audited web content.
Capability inventory: The skill possesses significant capabilities, including executing shell commands (pnpm, capacitor, vite), creating/modifying project files, and configuring build settings.
Sanitization: There is no evidence of sanitization or filtering of the external content before it is incorporated into the audit report and processed by the agent.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands for project scaffolding and Capacitor management.
Evidence: references/site-to-ios-runbook.md contains commands such as pnpm create vite site-ios --template react-ts, pnpm add @capacitor/core @capacitor/ios, and pnpm exec cap sync ios.
[EXTERNAL_DOWNLOADS]: The skill initiates downloads of standard software development packages during the scaffolding process.
Evidence: The use of pnpm add and pnpm create triggers downloads from the NPM registry. These target well-known services and are standard for the stated purpose of the skill.

site-to-ios-app