$ modelswap web
Every key. Every model. One plane.
Key and model control plane for AI agents: 41 provider presets out of the box, adapters for 10 agents, 37 usage sources queried directly, and multi-device sync (LAN peer-to-peer + self-hosted cloud). Switches only rewrite the fields ModelSwap owns — hooks and statusLine stay untouched, with automatic snapshots you can roll back — plus a CLI and Skill that other agents can call directly.
Claude Code
ChatGPT
Kimi
Hermes
WorkBuddy
Zhipu Coding Plan
MiniMax Coding
Claude
5h
8%
GLM
week
62%
DeepSeek
balance
81%
Two ways, up and running in a minute.
Desktop and CLI share the same data directory — pick either, or run both.
npm install -g modelswap
modelswap webAgent Config
Adapts to 10 agents: Claude Code, ChatGPT Codex, Kimi Code, WorkBuddy, Hermes and more. One click switches the model — config.toml and auth.json are written correctly, never hand-edited.
Claude Code
ChatGPT
Kimi
Hermes
WorkBuddy
Zhipu Coding Plan
MiniMax Coding
Built-in Providers
41 provider presets covering international, aggregator, and Chinese platforms — with Coding Plan, Token Plan and other subscriptions managed in one place.
Encrypted Vault
AES-256-GCM local encryption; keys bind to projects and inject into .env. The browser extension creates keys on provider consoles and files them back automatically — plaintext never hits disk.
sk-…•••7f2a → ~/.modelswap/vault (encrypted)
Usage Tracking
Live subscription and balance queries across 37 sources with smart polling; banner alerts and local notifications fire before quota runs out.
Claude
5h
8%
GLM
week
62%
DeepSeek
balance
81%
Config Safety
Surgical writes: ModelSwap only rewrites the fields it owns — your hooks, statusLine and MCP config stay untouched. Every switch takes an automatic snapshot you can compare and restore in one click.
Multi-device Sync
LAN peer-to-peer, or plug in your own cloud storage (Cloudflare / Supabase / WebDAV / iCloud and more — 9 backends); payloads stay encrypted so the server never sees plaintext, plus one-time sync codes to migrate between machines.
CLI & Skill
Every ModelSwap capability is callable: other agents create keys and switch models via the CLI or Skill — no dashboard required.
# Let your agent drive ModelSwap
modelswap web # open the dashboard
modelswap vault # key management
# Skill (coming soon) — natural language calls
"Create a new Zhipu key and store it"
"Switch Codex to GLM-5"Compared to the alternatives
Frequently asked
Where are my keys stored?
Keys live in an AES-256-GCM encrypted vault under ~/.modelswap on your machine. Cloud sync is end-to-end encrypted — plaintext never leaves your device.
Which agents are supported?
Ten built-in adapters: Claude Code, ChatGPT Codex, OpenCode, OpenClaw, WorkBuddy, ZCode, Hermes, Kimi Code, Grok and MiMo Code — each writes that agent’s native config files.
How does auto-create work?
The browser extension operates inside your provider console: it fills the create form, copies the new key, and files it into the ModelSwap vault — entirely between your browser and your machine.
Can other agents call ModelSwap?
Yes. The CLI is available today, and a built-in Agent Skill ships with it — run modelswap skill to install SKILL.md into your project’s .agents/skills/modelswap/ directory for any Agent-Skill-aware tool to use.
Will it conflict with existing configs?
No. ModelSwap only rewrites the fields it manages — your hooks, statusLine and MCP config stay untouched. Every switch also saves an automatic config snapshot (last 10 kept per agent) that you can compare and restore with one click in Settings, so you can always go back to official or the original provider.