The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data that could be used to manipulate agent behavior if it contains embedded instructions.
Ingestion points: Processes git commit messages via git log, user-supplied bullet points, conversational free text, and local JSON metadata files from the ./metadata directory.
Boundary markers: The instructions lack explicit delimiters or "ignore" directives to ensure the agent treats external data strictly as data rather than instructions.
Capability inventory: The skill utilizes shell commands (git and asc) to read repository data and upload metadata to App Store Connect.
Sanitization: There is no evidence of input validation or sanitization for the text extracted from git logs or user inputs before it is used to generate final drafts.
[COMMAND_EXECUTION]: The skill is built around the execution of shell commands to interact with the local environment and the App Store Connect service.
Evidence: Uses git describe and git log to analyze the project history.
Evidence: Uses the asc CLI utility for authentication (asc auth login), metadata synchronization (asc metadata pull/push), and direct app info updates (asc apps info edit). While these are the core intended functions, they define the skill's capability to modify external configurations.

asc-whats-new-writer