The Agent Skills Directory

Local Script Execution: The skill instructs the agent to run a repository-specific script eng/mark-shipped.ps1. Executing local scripts is a common pattern for automating project tasks, but it assumes the scripts within the repository are trusted and have been reviewed according to project standards.
Command Execution via CLI: The skill utilizes git and the GitHub CLI (gh) to perform actions like pushing tags and creating draft releases. These operations leverage the environment's authenticated credentials to modify remote repository state.
Indirect Prompt Injection Surface: The skill processes external data that could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: The skill reads a branch name from a user-provided issue description (Step 1) and ingests changelog content generated by the gh api command, which is derived from external commit messages or PR titles (Step 4).
Boundary markers: No explicit delimiters or instructions are provided to sanitize or ignore potential instructions embedded within the ingested data.
Capability inventory: The agent has capabilities to push to the remote repository, create releases, modify project files, and execute PowerShell scripts.
Sanitization: The version variable used in most shell commands is derived from a local configuration file (eng/Versions.props), which serves as a validation step that limits the data used in command arguments.

post-release-activities