ref-sp-agents-instructions-authoringlisted

# Agents Instructions Authoring ## Purpose Provide portable defaults for designing maintainable repository instruction systems across the major agent entry points without duplicating the same workflow, policy, and routing text in every provider file. ## When to use this skill - Creating or refactoring repo instruction files. - Deciding where the source of truth for instructions should live. - Adding Gemini or Claude support to a repo that already has Copilot instructions, or the reverse. - Reviewing whether top-level instruction files still match the codebase and the skill catalog. ## Scope Boundaries - Use this skill for the overall instruction architecture across providers. - Use `./references/providers/copilot-instructions.md` for what belongs specifically in `.github/copilot-instructions.md`. - Use `./references/providers/gemini-instructions.md` for `GEMINI.md` bridge or provider-specific decisions. - Use `./references/providers/claude-instructions.md` for `.claude/CLAUDE.md` bridge or provider-specific decisions. - Use `./references/global-instructions.md` for user-level/global instruction files (e.g. `~/.copilot/instructions/*.instructions.md`, `~/.claude/CLAUDE.md`) rather than repo-scoped files. - Use `./references/agents-md-standard.md` for the cross-provider root `AGENTS.md` file and how it fits the source-of-truth model. - Use the repo's agent-persona skill (`ref-sp-agents-mr-wolf-persona` here) when the instruction system needs to preserve or refresh the rep