shunk031-manage-public-private-dotfiles
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill is designed to interact with sensitive file paths including
~/.local/share/chezmoi-private,~/.config/chezmoi-private/chezmoi.yaml, and~/.agents/AGENTS-private.md. These paths are expected to contain credentials and internal infrastructure settings. The skill includes a mitigating instruction to keep secret values out of command output and chat summaries. - [PROMPT_INJECTION]: The skill contains a directive for the agent to output a specific phrase ("🏠 I read shunk031-manage-public-private-dotfiles") upon loading the SKILL.md, which is a minor instruction override.
- [INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to read and follow rules defined in external files, creating an indirect prompt injection surface.
- Ingestion points: The agent reads
~/.agents/AGENTS.md,~/.agents/AGENTS-private.md, and the rootAGENTS.mdwithin the dotfiles repositories. - Boundary markers: None are defined to distinguish between data and instructions within these files.
- Capability inventory: The skill uses
git statusandchezmoi source-path, and performs file modifications to the chezmoi source state. - Sanitization: There is no explicit sanitization or validation of the content read from the instructions files.
Audit Metadata