The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to automate git workflows, including branching, committing, and pushing changes. It also executes a project-specific script (pnpm bump). These operations are essential for the release process and are scoped to the local repository.
[PROMPT_INJECTION]: The skill ingests untrusted data from git commit logs and the changelog/next.md file. This data is processed by the agent to categorize changes, which introduces a surface for indirect prompt injection if commit messages contain malicious instructions.
Ingestion points: Git log output and apps/desktop/changelog/next.md file contents.
Boundary markers: No specific delimiters are implemented to separate external data from the agent's internal instructions.
Capability inventory: The skill allows file writing, git repository modifications, and execution of the local pnpm bump command.
Sanitization: There is no evidence of sanitization or filtering for the commit messages before they are presented for categorization.

desktop-release