Back to Trigr

Build in Trigr from your AI assistant

Connect ChatGPT, Claude, Codex, Claude Code, or another remote MCP client to Trigr’s catalog and your owner-scoped account context. It can research, build, backtest, optimize, save drafts, and manage your work while you monitor the results in Trigr.

1. Connect your client

Choose your client and add Trigr's remote MCP endpoint. No API key or environment variable is needed.

Remote MCP URL
https://trigr.xyz/mcp

2. Authorize in your browser

On ChatGPT web, enable Developer mode, create an app, paste the Trigr MCP URL, choose OAuth, and scan its tools.

  1. 1.Open ChatGPT on the web and turn on Developer mode in Settings → Security and login. Workspace plans may need an admin to allow it.
  2. 2.Go to ChatGPT Plugins, select +, and create a developer-mode app with the Trigr MCP URL and OAuth.
  3. 3.Complete the Trigr browser authorization, then enable the new Trigr app in a chat.
OpenAI documents developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web. Menus change often; follow OpenAI's developer mode guide if these steps differ.
Sign in to Trigr, review the permissions, and select Authorize connection. The revocable MCP credential is created automatically. MCP access is available on every Trigr plan; your client’s MCP availability, saved-strategy limits, and credit limits still apply.

3. Choose how autonomous it should be

Choose manual or autonomous mode during browser authorization. You can change the mode separately for every connected client.

4. Start a workflow

The starter workflow checks the account before proposing work. Ask naturally in any client; Claude Code also exposes MCP prompts as slash commands.

Start building
Use the Trigr MCP to build a strategy. First check my account context, existing strategies, and agents, then ask what I want to build.
Build

Design a new strategy from the supported catalog.

Improve

Review a saved strategy and its agent exposure.

Research

Compare marketplace candidates with your account.

Context and control

The MCP server gives the agent enough context to make relevant suggestions without exposing account secrets.

What it can inspect
  • Plan, subscription status, credits, points, and quotas
  • Saved strategies and optional graph recipes
  • Trading-agent summaries and linked strategy slots
  • Catalog, marketplace metadata, and exact data coverage
  • Create, rename, update, or delete owned strategies and edit agents
What stays protected
  • No API secrets, signing keys, or exchange credentials
  • No private marketplace recipes
  • Manual mode asks before each exact action; autonomous mode can be revoked at any time
  • No live trading, order placement, or withdrawals — publishing and new paper agents always use a single-use, exact-action authorization