The Agent Skills Directory

[Command Execution]: The skill executes various development-related shell commands including pip install, pytest, flake8, and clang-format-18. These are standard tools for CI/CD and local development workflows, used here to verify code quality and package integrity.
[Indirect Prompt Injection Surface]: The skill gathers context by reading git logs and file diffs. This is a potential surface where malicious instructions embedded in commit messages or code comments could influence the agent's summarization or documentation tasks.
Ingestion points: git log, git diff, CHANGELOG.md, pyproject.toml, and local plan files in .copilot/plans/ (specified in SKILL.md).
Boundary markers: No explicit delimiters are used to isolate external data from the skill's internal instructions.
Capability inventory: The skill can execute shell commands (pip, pytest, flake8), modify local project files (versioning and documentation), and write log files to .copilot/finalize/.
Sanitization: No specific sanitization or escaping of the data ingested from git or project files is described.
[External Package Installation]: The instructions suggest installing clang-format-18 using a system package manager (apt). This involves downloading and installing software from external repositories, which is a standard procedure for maintaining development environments.
[Runtime Compilation]: The skill performs an editable installation of the local package with BOCPY_BUILD_INTERNAL_TESTS=1. This environment configuration triggers the compilation of C extensions at runtime to ensure the test suite exercises the latest build.

finalize-pr