app-store-connect

Pass

Audited by Gen Agent Trust Hub on Nov 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the App Store Connect API that is then processed by the agent, creating a surface for indirect prompt injection.
  • Ingestion points: Customer reviews are fetched via GET requests to /v1/apps/{APP_ID}/customerReviews as described in references/reviews-and-ratings.md.
  • Boundary markers: Absent. The instructions do not provide delimiters or warnings to ignore instructions embedded within the fetched review text.
  • Capability inventory: The skill possesses the ability to make further network requests (requests), write files (reports-and-analytics.md), and execute shell commands (xcodebuild in build-and-deploy.md).
  • Sanitization: Absent. The workflow in references/reviews-and-ratings.md suggests using the agent to identify issues and draft replies based on review content without sanitizing the input.
  • [EXTERNAL_DOWNLOADS]: The skill documentation requires the installation of standard Python libraries from a public registry.
  • Evidence: The README.md instructs users to run pip install PyJWT requests and pip install "qrcode[pil]" Pillow. These are well-known, legitimate packages for handling JWT authentication, HTTP requests, and image/QR code generation.
  • [COMMAND_EXECUTION]: The skill utilizes system CLI tools to perform app builds and deployment tasks.
  • Evidence: references/build-and-deploy.md and references/screenshots-and-previews.md provide instructions for executing xcodebuild, xcrun simctl, and hashlib operations. These are standard tools for the stated purpose of iOS app management.
  • [DATA_EXFILTRATION]: The skill communicates with external domains to perform its functions.
  • Evidence: Network operations are directed at api.appstoreconnect.apple.com. As this is the official API for a well-known service (Apple), these connections are documented as part of the intended functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 2, 2026, 12:04 PM