Skip to content
Back to skills

Setup

ASecurity

Verify the tvremix MCP connection and tour the available commands

  • 13 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added May 26, 2026
ai-agentsgogitapi

Works with

  • claude code
  • claude desktop
  • cursor
  • cli
  • api
  • mcp

Security analysis

A100/100

Scanned May 27, 2026

npx -y skills add tvremix/claude-plugin --skill setup --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Setup?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Setup
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/tvremix-setup/badge)](https://www.skillsdirectory.com/skills/tvremix-setup)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: setup
description: Verify the tvremix MCP connection and tour the available commands
---

Walk the user through confirming their `tvremix` plugin is working and introduce them to the available commands. Keep the tone conversational.

## Step 1 — Confirm the environment

Tell the user you'll run a harmless smoke test. If they're seeing this skill, Claude Code is clearly running, so just confirm briefly.

## Step 2 — Authenticate (first run only)

This plugin connects to one MCP server:

- **tvremix** — `https://tvremix.xyz/api/mcp/v1` (TradingView data + SMC/swing analyzers + options + web search)

On the very first tool call, a browser window will open at `tvremix.xyz/oauth/authorize`. Tell the user:

1. Sign in with Google (or email OTP).
2. Click **Allow**.
3. You'll be redirected back and the tool will complete automatically.

If they've done this before, nothing opens — the stored refresh token is used silently.

## Step 3 — Smoke test

Call `mcp__tvremix__get_quote` with `symbol="NASDAQ:AAPL"`. This is the cheapest, safest call — public data, no side effects. Show the user the resulting price + change% + volume as a single line, e.g.:

> **AAPL** — $185.23 (+1.2%) on 41.8M shares — tvremix MCP is live.

If the call fails:
- `401 Unauthorized` → OAuth didn't complete. Ask the user to re-run the command to re-trigger the browser flow.
- `429 Too Many Requests` → rate limit (60/min/account). Wait a minute and retry.
- Any other error → report the full message and suggest they file an issue at `https://github.com/tvremix/claude-plugin/issues`.

## Step 4 — Quick tour

Tell the user about the available slash commands. All are prefixed with `/tvremix:`:

| Command | What it does | Example |
|---|---|---|
| `/tvremix:setup` | This verification skill | `/tvremix:setup` |
| `/tvremix:smc` | Smart Money Concepts analysis (BOS, CHoCH, OBs, FVGs, bias) | `/tvremix:smc NASDAQ:NVDA 4h` |
| `/tvremix:swing` | Swing trading setup with Fibonacci, pivots, R:R | `/tvremix:swing NASDAQ:AAPL` |
| `/tvremix:momentum` | Multi-timeframe RSI/MACD/volume momentum read | `/tvremix:momentum BINANCE:BTCUSDT` |
| `/tvremix:options` | Options chain, Greeks, strategy suggestions | `/tvremix:options NASDAQ:TSLA` |
| `/tvremix:levels` | Key S/R, order blocks, FVGs, liquidity levels | `/tvremix:levels NASDAQ:META` |
| `/tvremix:confluence` | Multi-timeframe trend alignment check | `/tvremix:confluence NASDAQ:MSFT` |
| `/tvremix:chart` | Quick price-action read (quote + bars + trend) | `/tvremix:chart NYSE:JPM` |

They can also just ask Claude anything about a symbol — the commands are shortcuts for common workflows.

## Step 5 — What's next

Suggest they try `/tvremix:smc NASDAQ:NVDA 4h` as a first real test — it exercises the most sophisticated tool (`analyze_smc_tool`) and produces a visually interesting structured read.

Remind them:
- Not a Claude Code user? The same tools are available to Claude Desktop, Claude.ai web, Cursor, OpenClaw, Hermes, and any other MCP client — see `https://tvremix.xyz/mcp`.
- Need API keys for CLI tools? `https://tvremix.xyz/account#api-keys`.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.