The Agent Skills Directory

[COMMAND_EXECUTION]: The Companion check section in SKILL.md instructs the agent to execute an ls command targeting several directories, including hidden folders within the user's home directory (e.g., ~/.claude/skills/). This activity constitutes local environment scanning and software discovery without explicit user initiation.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for the agent to recommend and facilitate the installation of additional tools from an external repository (AbsolutelySkilled/AbsolutelySkilled). This source is not verified as a trusted vendor and the repository name does not match the provided author context, presenting a risk of installing unverified code.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to process untrusted data from Git repositories (commit messages, logs, and diffs).
Ingestion points: Git log output, commit messages, and file diffs processed during rebase and bisect operations.
Boundary markers: None identified in the instructions to separate repository data from agent instructions.
Capability inventory: Subprocess execution for git, npm, and npx commands, as well as file system read/write access via git operations.
Sanitization: No explicit sanitization or validation of repository-derived strings is documented.

git-advanced