The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash to execute git commands (e.g., git ls-files, git rev-list, git cat-file) and standard Unix utilities (du, find, sha1sum, awk) for repository analysis. These commands are scoped to read-only diagnostics of file sizes and history.
[DATA_EXFILTRATION]: The skill includes logic to detect sensitive file paths such as .pem, .key, .p12, and id_rsa within the repository. This is an auditing feature intended to identify accidentally leaked credentials and does not include any network exfiltration or external access.
[PROMPT_INJECTION]: The skill processes untrusted repository metadata and file structure, which represents an indirect prompt injection surface. This is a known risk for file analysis tools but is categorized as safe here due to the skill's diagnostic nature and restricted local scope.
Ingestion points: git ls-files, git rev-list, find, and git cat-file in SKILL.md.
Boundary markers: None present.
Capability inventory: Read-only Bash commands for repository inspection.
Sanitization: None present.

manage-assets