update-instructions

To sync AI instruction rules across platforms, this skill delegates targeted rule updates to an instruction-engineer subagent, applying a standing-default preamble to ensure consistent policy behavior across Claude Code, Codex, and Gemini CLI.

The safest way to update global AI instruction files is through surgical rule edits. This approach avoids bulk replacement or scanning entire files, ensuring full auditability by only applying targeted changes via a specialized subagent.

You can apply a policy edit across multiple AI platforms by prepending a standing-global preamble and executing surgical updates. This prevents full file rewrites by routing changes through an instruction-engineer subagent for multi-step validation.

Yes, you can update Claude Code and Gemini CLI instruction files simultaneously. The skill synchronizes cross-platform changes by delegating the edits to an instruction-engineer subagent while enforcing standing defaults across all targets.

Using a subagent to edit AI instruction rules prevents direct edits in the main chat, ensuring prompt safety and full auditability. This delegation enforces multi-step validation and consistent cross-platform synchronization.

Yes, updating AI instruction rules with this approach is designed specifically for partial rule changes. It focuses on surgical edits rather than drafting or scanning entire instruction files, ensuring precise and auditable policy updates.