The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation instructs the user to install the asc CLI tool via Homebrew from the author's tap (tddworks/tap/asccli). This involves downloading and installing external executable code onto the system.
[DATA_EXFILTRATION]: The skill provides examples for authenticating the CLI tool using sensitive API credentials and private key files (e.g., ~/.asc/AuthKey_XXXXXX.p8). While these are necessary for the tool's operation, they represent access to highly sensitive security assets.
[COMMAND_EXECUTION]: The skill's primary functionality is achieved by executing the asc command-line utility with various arguments to manage App Store resources.
[PROMPT_INJECTION]: The skill instructions define an 'affordances' mechanism where the agent is encouraged to execute commands provided directly in the tool's JSON output. This creates a surface for Indirect Prompt Injection, as data retrieved from external App Store Connect APIs could theoretically contain malicious command strings that the agent would then execute.
Ingestion points: JSON responses from asc CLI commands (e.g., asc apps list) containing an affordances field.
Boundary markers: Absent; the agent is explicitly told to 'Always use these instead of constructing commands manually'.
Capability inventory: The asc tool can perform a wide range of actions including listing/creating versions, uploading builds, and managing TestFlight users.
Sanitization: There is no evidence of sanitization or verification performed on the commands retrieved from the tool's output before they are executed.

asc-cli