# MCP Config Sync

> Provider-neutral entry point for discovering, planning and synchronizing MCP configuration between user-selected providers and app classes. The user selects truth, targets and scope; no provider is an implicit hub.

- Skill: `ellmos-ai/mcp-config-sync` (Agent Skill, multi-file: 16 files)
- Install (CLI): `npx skillmds@latest add ellmos-ai/mcp-config-sync`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ellmos-ai/mcp-config-sync/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: ellmos-ai (https://skillmd.com/u/ellmos-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ellmos-ai/mcp-config-sync

---


<img src="banner.png" width="100%" alt="mcp-config-sync banner">

> **Deutsch** — Offizielle Deutsch-Version / Documento Oficial en Deutsch.


# MCP Config Sync (Deutsch)

This is the MCP-focused entry point to `agent-config-sync`. It assumes no
provider, app or master file.

1. Ask which concrete endpoints or axes the user wants: within one provider
   across app classes, within one app class across providers, an explicit list,
   or every detected provider and class.
2. Run `agent-config-sync/scripts/sync.py --discover`, then `--offer`.
3. Present detected endpoints separately from unverified candidates.
4. Let the user choose truth source, targets, direction and conflict policy.
5. Materialize `registry.json`, review `--plan`, and only then use
   `--apply --yes`.

Discovery and offers are read-only. There is no implicit hub and no implicit
“sync all”. The former Claude Code↔Claude Desktop scripts are a legacy profile,
not the generic default.