The Agent Skills Directory

[DYNAMIC_EXECUTION]: The checker utility evaluates the project's Expo configuration files at runtime to verify rule compliance. This involves dynamic loading of potentially untrusted local code.
Evidence: scripts/lib/evaluate-app-config.mjs uses await import(pathToFileURL(configPath).href) to load the configuration file.
Evidence: scripts/lib/app-config.mjs spawns a child process to execute the evaluation logic in a fresh environment.
[INDIRECT_PROMPT_INJECTION]: The skill operates as an analysis tool that ingests project-specific data which could contain instructions intended to influence the agent's behavior during the audit.
Ingestion points: app.config.ts, eas.json, and package.json in the target project.
Boundary markers: No explicit boundary markers or "ignore" instructions are used when interpolating evaluated data.
Capability inventory: The skill performs file system reads, executes shell commands (EAS CLI and Git), and performs dynamic JavaScript execution via child processes.
Sanitization: The configuration evaluator implements basic sanitization by converting function objects into placeholder strings during JSON serialization of the results.

ota-updates