Frase.io MCP Server

Frase for AI agents

Run the whole content pipeline from your own stack

Connect Frase over MCP, and the AI agents you already use, like Claude, Cursor, and Windsurf, drive the full Frase loop: research, write, optimize, publish, and monitor. Prefer the terminal? Run the same loop with the Frase CLI. One prompt does the work a whole stack used to.

Start free trialSee where AI ignores you

7-day free trialNo credit cardAPI on every plan

Trusted by content and SEO teams at

Connect once

Your agent calls Frase in plain language

Connect the Frase MCP server and your assistant can ask Frase to check a SERP, build a brief, draft a page, or score it, and read the answer back as structured data. Frase proposes the next step; you approve what runs.

• The full Frase tool surface, available the moment you connect
• Results come back as data your agent reads, not a dashboard
• Frase proposes — nothing publishes until you say so

The content pipeline

One prompt, the whole pipeline

Ask once, and Frase runs every stage in order, carrying the brief, the brand voice, and the scores from one step to the next. No tab-switching, no context lost between tools.

Research

Frase reads the SERP, the intent, and the gaps before a word is written.

Brief

Frase turns the research into a brief with the questions and structure to win.

Draft

Frase drafts in your brand voice from the brief, ready for a human read.

Optimize

Frase scores the draft for search and AI answers, then tightens it.

Publish

Frase publishes to WordPress, Webflow, Sanity, Wix, or FraseCMS.

Monitor

Frase tracks where AI engines cite the page and flags when it slips.

Every stage runs from a single prompt, the CLI, or one API call. Quality gates run before anything publishes. Frase ships content when it passes the bar you set, not when someone remembers to click.

Run it from the terminal

The whole content platform, from your terminal

The Frase CLI takes you from research to a published, optimized page in commands you can script. Every command has a JSON flag, so it drops straight into CI, cron jobs, and your own pipelines and feeds the next step automatically. The same API powers it all, so you can call it directly when you need to, with webhooks that fire the moment a long job is done.

• Research, draft, score, and publish in scriptable commands
• A JSON flag on every command, built for CI and automation
• The same API underneath, with webhooks that close the loop, no polling needed

MCP and CLI first

Reach Frase however your agents work

Lead with the MCP server inside your assistant or the CLI in your terminal. That's where the advantage is. The API and pre-built skills are there when you need them. Whichever you pick, it's the same Frase tool surface behind one key.

MCP server

Lead with this

Connect @frase/mcp-server and the full Frase tool surface lands inside Claude Desktop, Claude Code, Cursor, Windsurf, and any MCP-compatible assistant. Your agent runs research, writing, optimization, publishing, and AI-visibility checks in plain language. That's the whole loop, from the chat you already work in.

claude mcp add frase

CLI

Built for automation

Run the same loop from your terminal. The Frase CLI takes you from research to a published, optimized page in commands you can script. Every one carries a JSON flag, so it drops into CI, cron jobs, and your own pipelines and feeds the next step automatically.

npx @frase/cli

REST API

Under the MCP server and CLI sits one API. Call it directly to wire Frase into any custom workflow: research, brief, draft, score, publish, and read AI-visibility data over HTTP, with webhooks that fire the moment a long job is done.

next.frase.io/api/docs

Agent skills

Install @frase/agent-skills and your assistant learns the right Frase workflow: not just which tools exist, but when to research, when to optimize, and when to publish.

@frase/agent-skills

Close the loop

Your agent can read where AI cites you

After a page ships, Frase keeps watching. It tracks share of voice across ChatGPT, Perplexity, Gemini, and Google AI, and surfaces alerts when your visibility moves, so your agent knows which pages to revisit next. AI Visibility monitors and tells you; you decide what to act on.

• Daily share-of-voice across the AI engines that answer your buyers
• Alerts and insights your agent reads back as data
• The signal that tells the pipeline which page to improve next

AI Visibility Tracking daily

Share of voice for “best CRM software” across AI search

34%your share

• Vantrayou34%
• Larkfield26%
• Brightwall22%
• Tidewell18%

Tracked across

ChatGPT Perplexity Claude Gemini Google AI

Cited in 18 of 40 AI answers this week, up from 11.

AutonomyTrust, earned per type

FAQ pagesEarned

Consistently clears your bar. Now publishing on its own to FraseCMS.

Auto-publish: on

Glossary termsBuilding trust

Comparison pagesIn review

Every type starts in review. Nothing publishes on its own until you say so.

You stay in control

Hand off more as Frase earns your trust

Every content type starts in review. As you approve Frase's drafts, it tracks how often you say yes. Once a type clears your bar, you can let it auto-publish. Set a guardrail and Frase pauses on its own if quality slips. The autonomy is yours to grant, one step at a time.

• Trust is earned per content type, not switched on all at once
• Enable auto-publish only where you've seen the work hold up
• Frase pauses below the approval rate you set — a guardrail, not a gamble

Stop stitching point tools

One surface beats a duct-taped stack

Wiring an agent into five separate tools means five APIs, five sets of glue code, and context dropped at every handoff. Frase gives your agent the whole pipeline behind one key.

The fragmented way

• A research tool, a writer, an optimizer, a publisher, and a tracker. Five APIs to learn.
• Your agent hands off between tools, and context drops at every seam.
• Glue code to stitch dashboards your agent can't read.
• Publishing lives in a separate app you copy-paste into.

The Frase way

• One API key, one tool surface: research, write, optimize, publish, and monitor from the same place.
• Frase carries the brief, the brand voice, and the scores through the whole pipeline.
• Structured data your agent reads back, so the next step runs on the last step's result.
• Frase publishes to WordPress, Webflow, Sanity, Wix, or FraseCMS from the same call.

Prefer not to wire it up?

Frase has an agent built in

Everything on this page reaches Frase from your own tools. If you'd rather skip the setup, the Frase Agent lives inside the app. Brief it in conversation and it does the work there.

Meet the Frase Agent

Proof

Teams run their pipeline on Frase

4.8/5 on G2500+ reviews across G2 & Capterra

Thousands of content and SEO teams run the loop on Frase, from one prompt or their own pipeline.

What you get in the trial

Connect Frase and run a real pipeline today

In the trial you connect Frase to your assistant, run the pipeline on a real topic, and watch a researched, scored draft come back, then publish it to your CMS. You'll see the workflow before you pay for it.

The MCP server, CLI, and API are included on paid plans, with no enterprise tier required. Plans start with a 7-day free trial, and yearly saves 20%.

Start free trialSee plans

7-day free trial. No credit card.

Frequently asked questions

Which AI tools can call Frase?

Frase works with any MCP-compatible assistant, including Claude Desktop, Claude Code, Cursor, and Windsurf, through the @frase/mcp-server package. You can also reach the same capabilities from the Frase CLI in your own scripts and CI, or from the REST API in any custom workflow. One API key connects all of them.

What can my agent actually do through Frase?

The full Frase content pipeline: research a topic and the SERP, build a brief, draft in your brand voice, score the draft for search and AI answers, and publish to your CMS. Your agent can also read AI-visibility data to see where AI engines cite your pages. It's the same capabilities the Frase app offers, callable from your stack.

Does Frase publish content on its own?

Only when you opt in. By default Frase proposes each step and waits for your approval, so nothing publishes unasked. For content types you trust, you can enable auto-publish to a connected CMS, and Frase pauses it automatically if quality drops below the bar you set. You stay in control.

Where can my agent publish?

Frase publishes to WordPress, Webflow, Sanity, Wix, and FraseCMS. You connect a CMS once, and your agent can publish there as the last step of the pipeline.

How is this different from the agent inside Frase?

This page is about reaching Frase from the AI tools you already use: your assistant, your CLI, your code. The Frase Agent is the assistant built into the Frase app, where you brief it in conversation and it does the work there. Same capabilities, two front doors: bring your own agent here, or use the one inside Frase.

Do I need an enterprise plan to use the MCP server, CLI, or API?

No. The MCP server, CLI, and API are available on paid plans starting with a 7-day free trial, with no enterprise tier required. White-label client portals and custom domains are the parts reserved for Enterprise.

Give your agent the whole pipeline

Connect Frase to your assistant, your CLI, or your code, and run research to published, monitored content from one prompt.

Frase MCP serverSee plansCheck your AI visibility

Start free for 7 days

No credit card required

Try Frase Free →

AI Content Pipeline | Run Frase From Your AI Agents | Frase.io