The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several powerful CLI tools, including xcrun simctl for simulator management, git and gh for repository and PR tasks, and npm/npx for running tests and linters. These commands allow the agent to modify the environment and the codebase significantly.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its ingestion of untrusted data from external sources.
Ingestion points: Processes testing workflows from /workflows/ios-workflows.md and UI/UX comparisons from WebSearch results.
Boundary markers: The skill lacks explicit delimiters or instructions to sub-agents to ignore potentially malicious directions within the ingested data.
Capability inventory: File system access, shell command execution, git operations, and simulator control.
Sanitization: No sanitization or validation of input data is performed before it is used to guide the agent's autonomous actions.
[REMOTE_CODE_EXECUTION]: The skill spawns autonomous sub-agents that are authorized to explore the codebase and implement code changes, which are then verified by executing test suites. This workflow could be exploited if a malicious workflow file leads the agent to generate and execute unauthorized code.
[EXTERNAL_DOWNLOADS]: The skill uses WebSearch to find and visit external websites for design research, exposing the agent to potentially malicious content on the web during its evaluation phase.

ios-workflow-executor