Provider-neutral planner for synchronizing MCP configuration, skills and rule files across agent providers and app classes. It discovers evidenced local options and lets the user choose truth, targets, direction and conflicts.
Scanned 9/4/2026
Install to Claude Code
npx -y skills add ellmos-ai/skills --skill agent-config-sync --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Agent Config Sync?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/ellmos-ai-agent-config-sync)More formats (shields.io, HTML) on the badges page.
---
name: agent-config-sync
version: 0.3.0
type: protocol
author: Lukas Geiger + Claude + Codex
created: 2026-06-20
updated: 2026-07-27
description: Provider-neutral planner for synchronizing MCP configuration, skills and rule files across agent providers and app classes. It discovers evidenced local options and lets the user choose truth, targets, direction and conflicts.
standalone: true
anthropic_compatible: true
bach_compatible: true
bach_origin: false
category: infrastructure
tags: [mcp, skills, rules, sync, provider-neutral, discovery, multi-agent]
language: de
status: active
visibility: public
aliases: [mcp-skill-sync, multi-agent-sync, tool-config-sync, agent-sync]
dependencies: {'tools': ['python'], 'services': [], 'protocols': [], 'python': []}
provenance: {'origin': 'custom', 'origin_path': 'skills/infrastructure/agent-config-sync/', 'origin_version': '0.3.0', 'last_sync_from_origin': None, 'last_sync_to_origin': None, 'local_changes_since_sync': False}
---
<img src="banner.png" width="100%" alt="agent-config-sync banner">
> **Deutsch** — Offizielle Deutsch-Version / Documento Oficial en Deutsch.
# Agent Config Sync (Deutsch)
The skill separates endpoint selection, resources and truth. Run:
```bash
python scripts/sync.py --discover
python scripts/sync.py --offer
```
The user can select an explicit endpoint list, one provider across app classes,
one app class across providers, or all detected endpoints. Detection is
evidence, not authorization.
Truth can be one endpoint, one file, an ordered set of files such as multiple
`AGENTS.md` layers, or a skills directory. No filename or provider is the
implicit hub. Without a selected truth source, plans stay blocked.
Review `--status` and `--plan`; only use `--apply --yes` after approval.
MCP blocks and skill directories are implemented. Rule-file topologies remain
fail-closed until the user selects a merge/redirect adapter.Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!