The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from git commit messages and GitHub pull request metadata (ingestion points identified in SKILL.md Steps 2 and 3).
Boundary markers: The skill instructs the agent to categorize content based on specific conventional commit prefixes, which provides a logical structure for data processing.
Capability inventory: The skill has the capability to modify the local CHANGELOG.md file and perform external actions via gh release create.
Sanitization: The skill implements a mandatory human review and approval step (Step 7 and Key Principle 1) before any filesystem writes or external API calls are executed. This human-in-the-loop requirement is a robust mitigation against malicious instructions embedded in repository history.
[COMMAND_EXECUTION]: The skill utilizes standard version control and repository management tools, specifically git and gh (GitHub CLI). The commands are used with specific, non-arbitrary arguments that align with the skill's stated purpose of release management.
[EXTERNAL_DOWNLOADS]: The skill fetches pull request and release information from GitHub. As GitHub is a well-known service and the retrieval is central to the skill's devops functionality, these operations are considered safe.

release-notes