ai-dev-tools-sync

name	ai-dev-tools-sync
version	1.0.0
description	[AI & Tools] Use when syncing supported AI dev-tool configuration, skills, prompts, agents, instructions, or workflows.

Quick Summary

Goal: Synchronize supported AI dev-tool configurations so Claude, Codex, and GitHub Copilot can work from equivalent source guidance where their surfaces overlap.

Workflow:

1. Understand — Read current configs (CLAUDE.md, AGENTS.md, CODEX_CONTEXT.md, copilot-instructions.md, agents, workflows)
2. Research — Search for latest features across the target platforms
3. Compare — Identify gaps in skills, prompts, agents, or instructions
4. Sync — Implement changes in source surfaces first, then regenerate managed mirrors where applicable

Key Rules:

• Copilot reads .claude/skills/ automatically (backward compatibility)
• Both platforms read .github/prompts/*.prompt.md and .github/agents/*.md
• Always update both CLAUDE.md and .github/copilot-instructions.md + .github/instructions/*.instructions.md for instruction changes
• Codex managed outputs are .agents/skills/**, .codex/CODEX_CONTEXT.md, and AGENTS.md; edit .claude source first and run Codex sync

Be skeptical. Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence percentages (Idea should be more than 80%).

AI Dev Tools Sync

Synchronize supported AI dev-tool configurations to maintain parity across active tool surfaces.

When to Use

Scope vs related skills: This is the broadest sync — bidirectional (Claude↔Copilot), covering skills, prompts, agents, and instructions. For Claude→Copilot knowledge/docs only → /sync-to-copilot. For the workflows.json catalog only (fast, no AI) → /sync-copilot-workflows.

Activate this skill when:

• User asks to update Claude Code or Copilot setup
• User wants to add/modify skills, prompts, agents, or instructions
• User wants both tools to work similarly
• User asks about AI dev tool configuration

Quick Reference

Source surface	Managed/peer surface	Location
SKILL.md	Codex skill mirror	.claude/skills/ -> .agents/skills/
Context/workflows	Codex context mirror	.claude/**, .claude/workflows.json -> .codex/, AGENTS.md
SKILL.md	Copilot skill/prompt	.claude/skills/ + .github/skills/ / .github/prompts/
agents/*.md	agents/*.md	.github/agents/ (shared)
CLAUDE.md	Copilot + Codex root instructions	Root CLAUDE.md, AGENTS.md, .github/
-	chatmodes/*.chatmode.md	.github/chatmodes/

Sync Process

Step 1: Understand Current Setup

Read these files to understand current configuration:

.claude/workflows/orchestration-protocol.md
.claude/workflows/primary-workflow.md
.github/copilot-instructions.md
.github/instructions/*.instructions.md
.github/AGENTS.md
CLAUDE.md

Step 2: Research Latest Features

Search web for:

• "GitHub Copilot features setup 2026"
• "GitHub Copilot custom instructions agents skills prompts"
• "GitHub Copilot agent mode workspace context"

See references/copilot-features.md for feature catalog.

Step 3: Identify Sync Opportunities

Compare capabilities and identify gaps:

• Skills missing in one platform
• Inconsistent prompt/instruction behavior
• Agent definitions that differ

Step 4: Implement Changes

For each change:

1. Skills: Create in both .claude/skills/ and .github/skills/
2. Prompts: Create in both .claude/skills/ and .github/prompts/
3. Instructions: Update CLAUDE.md + .github/copilot-instructions.md + .github/instructions/*.instructions.md
4. Agents: Update .github/agents/ (shared by both)

Compatibility Notes

• Copilot reads .claude/skills/ automatically (backward compatibility)
• Both read .github/prompts/*.prompt.md
• Both read .github/agents/*.md
• Both read AGENTS.md in root or .github/
• Both support path-based instruction files via applyTo in frontmatter

References

• Copilot Features Catalog
• Sync Patterns

---

[IMPORTANT] Use TaskCreate to break ALL work into small tasks BEFORE starting — including tasks for each file read. This prevents context loss from long files. For simple tasks, AI MUST ATTENTION ask user whether to skip.

AI Mistake Prevention — Failure modes to avoid on every task:

Check downstream references before deleting. Deleting components causes documentation and code staleness cascades. Map all referencing files before removal. Verify AI-generated content against actual code. AI hallucinates APIs, class names, and method signatures. Always grep to confirm existence before documenting or referencing. Trace full dependency chain after edits. Changing a definition misses downstream variables and consumers derived from it. Always trace the full chain. Trace ALL code paths when verifying correctness. Confirming code exists is not confirming it executes. Always trace early exits, error branches, and conditional skips — not just happy path. When debugging, ask "whose responsibility?" before fixing. Trace whether bug is in caller (wrong data) or callee (wrong handling). Fix at responsible layer — never patch symptom site. Assume existing values are intentional — ask WHY before changing. Before changing any constant, limit, flag, or pattern: read comments, check git blame, examine surrounding code. Verify ALL affected outputs, not just the first. Changes touching multiple stacks require verifying EVERY output. One green check is not all green checks. Holistic-first debugging — resist nearest-attention trap. When investigating any failure, list EVERY precondition first (config, env vars, DB names, endpoints, DI registrations, data preconditions), then verify each against evidence before forming any code-layer hypothesis. Surgical changes — apply the diff test. Bug fix: every changed line must trace directly to the bug. Don't restyle or improve adjacent code. Enhancement task: implement improvements AND announce them explicitly. Surface ambiguity before coding — don't pick silently. If request has multiple interpretations, present each with effort estimate and ask. Never assume all-records, file-based, or more complex path. Keep domain concepts out of generic/shared/infrastructure layers. A reusable layer (shared library, framework, infra module) must reference NO consumer-specific domain concept — tenant/customer/product IDs, business entities, feature rules. The leak compiles and runs, so it passes review silently while coupling the "reusable" layer to one consumer. Push domain fields/logic down into the consumer via subclass or composition.

Critical Thinking Mindset — Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence >80% to act. Anti-hallucination: Never present guess as fact — cite sources for every claim, admit uncertainty freely, self-check output for errors, cross-reference independently, stay skeptical of own confidence — certainty without evidence root of all hallucination.

MUST ATTENTION apply critical thinking — every claim needs traced proof, confidence >80% to act. Anti-hallucination: never present guess as fact.

MUST ATTENTION apply AI mistake prevention — holistic-first debugging, fix at responsible layer, surface ambiguity before coding, re-read files after compaction.

Closing Reminders

• MANDATORY IMPORTANT MUST ATTENTION break work into small todo tasks using TaskCreate BEFORE starting
• MANDATORY IMPORTANT MUST ATTENTION search codebase for 3+ similar patterns before creating new code
• MANDATORY IMPORTANT MUST ATTENTION cite file:line evidence for every claim (confidence >80% to act)
• MANDATORY IMPORTANT MUST ATTENTION add a final review todo task to verify work quality

[TASK-PLANNING] Before acting, analyze task scope and systematically break it into small todo tasks and sub-tasks using TaskCreate.