The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs extensive shell command execution to manage the repository lifecycle. Evidence includes the usage of git log, git push, gh release create, terraform validate, and uv run for executing local Python scripts. It also runs various local maintenance scripts such as scripts/check-github-release-controls.sh and scripts/validate_v3_final_polish_examples.py.
[PRIVILEGE_ESCALATION]: The skill is authorized to perform high-privilege operations on the repository's protected branch. Evidence includes the execution of git push --atomic --force-with-lease to the master branch and the creation of annotated tags.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could contain malicious instructions from third-party contributors. 1. Ingestion points: The agent reads CHANGELOG.md using awk and processes git log output containing commit messages from community contributors. 2. Boundary markers: None identified. The skill lacks explicit delimiters to distinguish data from instructions when processing extracted changelog entries. 3. Capability inventory: The skill has the ability to write to files, push to remote repositories, and execute local scripts. 4. Sanitization: No sanitization or validation of the content extracted from CHANGELOG.md or git log is performed before use in automation tasks.

prepare-release