lium-skill
MCP serverSearchSearch and read the Lium GPU rental docs: pods, CLI, SDK, REST API and agent guides.
Available today. Use it from your connected AI after setup.
No other account needed.
Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.
Then ask your AI: use lium-skill to search
Install lium-skill
The server’s own address, for the clients that take one directly. Or connect ahel once and every client you use reads it from one address, with the account kept on ahel rather than in each client’s config.
Claude Code
claude mcp add --transport http --scope user lium-skill 'https://docs.lium.io/mcp'Run it once in your project, then open /mcp to approve any sign-in the server asks for.
Claude Desktop
https://docs.lium.io/mcpAdd a custom connector in Settings, paste this address, and approve the sign-in.
Cursor
cursor://anysphere.cursor-deeplink/mcp/install?name=lium-skill&config=eyJ1cmwiOiJodHRwczovL2RvY3MubGl1bS5pby9tY3AifQ==Open the link and Cursor adds the server at that address.
ChatGPT
https://docs.lium.io/mcpIn Settings, enable Developer mode, create an MCP app, and paste this address. Your plan and workspace must allow custom apps.
Codex
codex mcp add lium-skill --url 'https://docs.lium.io/mcp'Run it once, then sign in with codex mcp login lium-skill if the server asks for an account.
From the project's README
As published by datura-ai/lium-skill in README.md.
The lium agent skill — teaches an AI coding agent to rent GPUs and manage pods
on Lium, agent-first compute: a GPU rental platform.
Install
npx skills add Datura-ai/lium-skill --skill lium
Works in any agent that supports the Agent Skills standard — Claude Code, Cursor, Codex and others.
Without node, install into the agent skill directories directly:
curl -fsSL https://lium.io/agents/install.sh | bash
The installer writes SKILL.md and references/ into ~/.claude/skills/lium/,
~/.cursor/skills/lium/ and ~/.codex/skills/lium/; it takes --claude-only,
--cursor-only, --codex-only, --force (overwrite without a backup) and
-h. LIUM_SKILL_VERSION is the git ref the files are fetched from (default
main); the repository has no tags yet, so pin a commit sha until one exists.
The skill drives the lium CLI, so install that too:
curl -fsSL https://lium.io/install.sh | bash
lium init
What is in here
| Path | Purpose |
|---|---|
lium/SKILL.md | The skill itself — frontmatter plus agent instructions |
lium/references/ | CLI and SDK references, loaded on demand |
agents/install.sh | Source of lium.io/agents/install.sh (lium.io redirects to the raw file on main) |
llms.txt | Short discovery index for agents |
llms-full.txt | Self-contained reference, generated from the files above |
scripts/build-llms-full.sh | Generates llms-full.txt (SKILL.md without frontmatter + both references) |
server.json | MCP Registry entry for the docs MCP endpoint https://docs.lium.io/mcp; a change on main publishes it (bump version) |
.github/CODEOWNERS, .github/REVIEWING.md | Who reviews each path and how |
lium.io/llms.txt and lium.io/llms-full.txt are served from this repo's main.
Editing
llms-full.txt is generated. After changing lium/SKILL.md or anything under
lium/references/, regenerate and commit it in the same change — the
check-drift job (.github/workflows/check-llms-full.yml)
regenerates the file on every PR, push to main and merge-queue run and fails
on any difference:
./scripts/build-llms-full.sh
Links
- Docs: https://docs.lium.io
- CLI: https://github.com/Datura-ai/lium
- PyPI: https://pypi.org/project/lium.io/
License
Tools it offers (2)
What this server listed when ahel dialed its public endpoint in Oct 2026, with no key and no account of yours. The names are the server’s own.
searchread_page
Signals
- Last commit
- Oct 2026
Advanced
- Delivery
- lium-docs MCP server → your ahel connector (mcp.ahel.ai) → your AI.
- Item type
- mcp-server
- Key
io-github-datura-ai-lium-docs- Source
- github.com/datura-ai/lium-skill
- Hosted endpoint
https://docs.lium.io/mcp
github.com/datura-ai/lium-skill