gauge-chat

Installation
SKILL.md

Gauge Chat

Be the answer for AI. Help the user's brand earn accurate, relevant recommendations when buyers ask ChatGPT and other chat or search models about their market. Measure what models say about the brand and its competitors, learn which sources and messages are working, and improve the content behind those answers so models become a marketing channel for the business.

Use a continuous loop: measure answers, understand the market, act on the evidence, and measure again. Route coding-agent tool selection and implementation evals to Gauge Agents instead.

Operate headlessly with Gauge MCP

Use the connected Gauge MCP tools for research and execution. Discover current tool descriptions and schemas before calling them; available integrations and operations depend on the connected account.

If Gauge is not connected, use the MCP setup guide. The server URL is https://app.withgauge.com/mcp. OAuth is the default for interactive clients; unattended clients can use an organization API key through an Authorization: Bearer header supplied by their secret configuration. Setup details are in Gauge under Settings → Integrations → MCP. Installing this skill does not establish that connection.

Resolve the organization and category before querying. For agency connections, discover the client organizations and use the intended client's scope. Read available company context, competitor context, content strategy, and writing guidance through the memory tools. Do not assume the MCP client already has the context used by Gauge's in-app agent.

Use direct tools for specific data. Use ask_gauge for an analytical question that benefits from synthesis, including the client, category, dates, filters, and required evidence. Pass relevant context explicitly on follow-ups. Use dedicated write tools for changes; do not assume a request made to the analyst saved anything.

Measure what models say

Start with the user's audience, product, competitors, and buying questions. Inspect tracked topics and prompts to establish what the data represents. Distinguish category discovery prompts from branded questions about the company; both are useful, but answer different questions.

Installs
3
First Seen
Sep 30, 2026