The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could serve as a vector for indirect prompt injection attacks.
Ingestion points: The skill reads canonical metadata from the ./metadata directory and routing app coverage data from ./coverage.geojson (SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore embedded instructions within these external files are provided in the skill text.
Capability inventory: The skill utilizes the asc CLI to perform significant mutations on App Store Connect, including staging versions (asc release stage), uploading artifacts (asc publish appstore), and submitting for review (asc review submit and asc review submissions-submit).
Sanitization: There is no evidence of content sanitization or validation of the ingested file contents prior to their use in release commands.
[COMMAND_EXECUTION]: The skill is designed to execute shell commands using the asc CLI tool. This grants the agent the capability to perform authenticated actions on the user's App Store Connect account, including creating submissions and modifying app metadata.

asc-release-flow