chezmoi-management

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 1.00). This skill's workflow explicitly runs run_after_apm-install.sh which performs "apm install --global --target claude" and references external APM-managed packages (dot_apm/apm.yml) that are fetched from public repositories (e.g., moonbitlang, mizchi, ast-grep), meaning untrusted third-party skill code/repos are downloaded into ~/.claude/skills and can influence the agent's behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). Yes — the skill runs chezmoi init https://github.com/mizchi/chezmoi-dotfiles.git (clone+apply) and then executes run_after_apm-install.sh which calls apm install to fetch/install external APM-managed skills into ~/.claude/skills, meaning remote repositories (starting with the flagged https://github.com/mizchi/chezmoi-dotfiles.git) are fetched at runtime and can deliver/execute code or agent skills that control prompts.