# Llmquant Setup

> Connect LLMQuant Data to this Hermes instance, confirm the DeepSeek baseline, smoke-test auth for zero credits, install the playbooks, and schedule the first standing job. Use when the user says "set up llmquant", "connect llmquant data", "llmquant setup", or otherwise asks to wire LLMQuant Data into Hermes.

- Skill: `llmquant/llmquant-setup` (Agent Skill)
- Install (CLI): `npx skillmds@latest add llmquant/llmquant-setup`
- Raw SKILL.md: https://api.skillmd.com/api/skills/llmquant/llmquant-setup/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: llmquant (https://skillmd.com/u/llmquant)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/llmquant/llmquant-setup

---


# LLMQuant Setup

Thin entry point for connecting LLMQuant Data to Hermes. **This skill does not
contain the setup steps** — the single canonical source is `SETUP.md` at the
root of this repo. Your job is to fetch it and follow it exactly, so this
skill never drifts out of sync with the real procedure.

## Do this

1. Fetch the canonical setup document:
   <https://raw.githubusercontent.com/LLMQuant/llmquant-hermes/master/SETUP.md>
2. Follow it top to bottom. It is agent-facing and self-contained: a trust
   contract, an instance self-check, DeepSeek baseline guidance, and setup
   steps (get key or Remote MCP URL -> write config -> smoke-test -> install
   playbooks -> schedule first cron), each ending in a check.
3. If this repo is already tapped locally, read the local `SETUP.md` instead
   of fetching — same document.

## If you cannot reach the network (fallback only)

The fetched `SETUP.md` is authoritative and far more complete. Use this
skeleton *only* if you truly cannot retrieve it, and tell the user you're
running from memory:

- **Never echo secrets.** Back up `~/.hermes/config.yaml` before editing;
  show the user the diff before writing; prefer the hosted Remote MCP URL.
- **Step 0:** self-check with `hermes --version`, `hermes skills tap --help`,
  `hermes cron --help`, `hermes model --help` — verify CLI syntax on this
  instance before using it.
- **Model:** DeepSeek is the recommended baseline; do not change model config
  without user consent.
- **Get credentials:** ask the user to sign up at <https://llmquantdata.com>
  and paste a Remote MCP URL (Dashboard → Connect) or API key (→ API Keys).
- **Write config:** add/update a single `llmquant-data` entry under
  `mcp_servers` (`url` preferred; stdio `command`/`args`/`env` fallback),
  then reload MCP.
- **Smoke-test:** call the 0-credit `personal_profile`; report
  `meta.remainingCredits`.
- **Install + schedule:** `hermes skills tap add LLMQuant/llmquant-hermes`,
  verify the skills are actually listed, then create one cron for the
  playbook, channel, and time the user picks.