The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several system commands including git, chezmoi, and a custom dotfiles helper to perform repository operations, file synchronization, and status checks.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from local and remote repositories, creating an attack surface for indirect prompt injection. 1. Ingestion points: Reads file contents and diffs from ~/.agents, ~/.local/share/chezmoi, and the $HOME directory as specified in SKILL.md. 2. Boundary markers: Uses an 'Enriched Report' phase (Phase 2) to present all classified changes for user review before execution (Phase 3). 3. Capability inventory: Includes capabilities to write to the home directory via chezmoi apply and push changes to remote Git repositories. 4. Sanitization: Explicitly instructs the agent to identify and exclude secrets, credentials, and raw credential diffs from all reports and commits, and requires semantic review of conflicts.
[PERSISTENCE]: The skill describes a recurring machine sync using a 'host-local hourly Codex tick' to maintain synchronization state across sessions.

dotfiles-sync