The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local Git commands including git status, git log, and git diff to audit the repository for task-related changes. These operations are scoped to the project environment and serve the intended purpose of the skill.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill reads project-specific documentation and task plans from the docs/ folder and records its findings in the advisor/ directory. It does not access sensitive user credentials or perform any external network requests.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an ingestion surface by parsing Markdown and YAML files from the docs/ directory. Ingestion points include document frontmatter and task titles (SKILL.md, verification-algorithm.md); sanitization is performed via a regex-based keyword extraction function; capability inventory is limited to local file and Git operations across the repository; and explicit boundary markers are absent. The data is used for keyword matching within a scoring algorithm.

verify-task-doc