The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting untrusted data from external sources.
Ingestion points: Commit messages are retrieved via git log and pull request details are fetched using gh pr view.
Boundary markers: The workflow does not explicitly define markers to isolate this untrusted content.
Capability inventory: The agent can perform sensitive actions such as git commit, git tag, and git push.
Sanitization: No explicit sanitization or validation of the commit/PR data is specified before it is used to generate changelogs. However, this surface is inherent to the skill's primary purpose of release automation.
[COMMAND_EXECUTION]: The skill utilizes several command-line tools to perform its tasks.
Tool list: It executes git for version control management and gh (GitHub CLI) for PR and repository interaction.
Context: These commands are used for legitimate, expected operations like analyzing history, committing changes, and pushing tags to remote repositories.

release-skills