The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands, primarily git and gh (GitHub CLI), to automate version control workflows. These commands are essential to the skill's purpose, including checking repository status, managing release branches, pushing commits, and creating pull requests.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection because it ingests and processes data that may contain malicious instructions (commit messages).
Ingestion points: The skill reads the commit history from the repository using git log {last_tag}..HEAD --oneline (in Step 3).
Boundary markers: The instructions do not provide explicit delimiters or instructions to the model to ignore potential commands within the commit messages when generating the CHANGELOG draft.
Capability inventory: The skill possesses capabilities that could be abused if an injection were successful, such as modifying source code (via the Edit tool), committing changes to the repository, and pushing to remote branches.
Sanitization: The skill implements a robust mitigation via a mandatory human-in-the-loop (HITL) step. In Step 3, the agent is instructed to show the CHANGELOG draft to the user and wait for explicit confirmation ([y/N/edit]) before proceeding to update files or commit changes.

publish