The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various shell commands to manage the environment, including git fetch, git status, git pull, and ln -s for symlink creation. It also uses command -v to detect installed tools and performs file operations like backing up files with a .bak extension.
[REMOTE_CODE_EXECUTION]: The skill utilizes npx skills update and npx skills add to manage agent skills. This involves executing a package runner to download and install code, primarily targeting the author's own 'rules-of-engagement' repository.
[EXTERNAL_DOWNLOADS]: The skill fetches updates and configurations from the mister-gnommer/rules-of-engagement GitHub repository to ensure the local machine stays in sync with the remote source.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to read content from the remote repository (e.g., core/codex.md, skills-list/skills.md) and local configuration files to update agent instructions. This creates a surface where instructions embedded in those sources could influence agent behavior.
Ingestion points: Remote repository files and local agent configurations (~/.claude/CLAUDE.md, ~/.codex/config.toml).
Boundary markers: The instructions explicitly require the agent to show a diff and obtain user approval before replacing or overwriting any files.
Capability inventory: Shell execution (git, npx, ln), and file read/write access to sensitive directories including ~/.claude/, ~/.config/opencode/, and ~/.agents/.
Sanitization: No automated sanitization is specified; the security model relies on the user's manual review of the proposed changes.

sync-setup