The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill repeatedly instructs the agent to use the command npx testflight as a shortcut for TestFlight submissions. This command downloads and executes an unverified third-party package from the NPM registry that is not an official Expo tool. The official deployment workflow is managed via the eas-cli package.
[METADATA_POISONING]: The skill's frontmatter claims the source is the official Expo repository (https://github.com/expo/skills/...) and labels the source type as official. However, the identified author is sickn33. This impersonation creates a deceptive trust profile.
[DYNAMIC_EXECUTION]: The reference file references/app-store-metadata.md encourages the use of store.config.js, which allows the execution of arbitrary JavaScript code during the build and deployment lifecycle. This enables potential runtime manipulation of application metadata and environment variables.
[EXTERNAL_DOWNLOADS]: The documentation provides code examples that fetch configuration data from an external, unknown domain (api.example.com). Using external network sources for deployment configuration introduces a significant supply chain risk.
[CREDENTIALS_UNSAFE]: The skill provides detailed instructions for managing and providing access to sensitive credentials, including Google Play Service Account JSON keys and Apple App Store Connect API keys (.p8 files). The management of these keys by the agent increases the risk of credential exposure.
[INDIRECT_PROMPT_INJECTION]: The dynamic configuration mechanism is susceptible to indirect prompt injection. If an attacker controls the external API used for metadata (e.g., in the api.example.com example), they can inject malicious content into the deployment pipeline.
Ingestion points: references/app-store-metadata.md (dynamic fetch calls in JS configuration).
Boundary markers: Absent in the provided examples.
Capability inventory: eas metadata:push and eas submit capabilities used to transmit data to app stores.
Sanitization: The skill does not demonstrate any validation or sanitization of the remote data before submission.

expo-deployment