# Copilot Review Instructions Sync > Update and validate GitHub Copilot review instruction files from Agent OS standards and repository conventions. Use when this capability is needed. - Skill: `tomevault-io/copilot-review-instructions-sync-2` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/copilot-review-instructions-sync-2` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/copilot-review-instructions-sync-2/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/copilot-review-instructions-sync-2 --- Use this workflow when standards change and Copilot review instructions need to stay in sync. ## Standards to read first - `agent-os/standards/index.yml` - Any changed standards under `agent-os/standards/**` - `AGENTS.md` ## Procedure 1. Inspect current Copilot instruction files: - `.github/copilot-instructions.md` - `.github/instructions/**/*.instructions.md` 2. Map standards to instruction scopes: - Repository-wide policy goes in `.github/copilot-instructions.md`. - Domain-specific policy goes in path-specific files with `applyTo` frontmatter. 3. Keep guidance concise and review-oriented: - Prioritize security, correctness, reliability, and performance. - Avoid style-only directives that create noisy reviews. 4. Ensure path-specific files have valid `applyTo` globs and no conflicting directives. 5. Validate practical review context is included: - CI parity commands (`bun run lint`, `bun run openapi:check`, `bun test`). - Contract-sync expectations for OpenAPI and SDK artifacts. 6. Enforce Copilot code review constraints: - Keep each custom instruction file under 4000 characters. 7. Report changes: - Which standards were reflected. - Which instruction files changed. - Any remaining standards not yet represented. ## Guardrails - Keep instructions declarative and broadly applicable. - Do not include secrets or environment-specific credentials. - Prefer standards as source of truth; only encode procedural guidance in skills. --- > Source: [ZebulonRouseFrantzich/chimera-bench](https://github.com/ZebulonRouseFrantzich/chimera-bench) — distributed by [TomeVault](https://tomevault.io).