Sign in Get Started
Model Context Protocol

RapidTextAI MCP Server

Give Claude, Cursor, or your own AI agent a direct tool for article generation. The RapidTextAI MCP Server exposes article writing, model discovery, and usage limits through the Model Context Protocol using JSON-RPC 2.0 over HTTP.

MCP server version 1.1.0. API access is available on paid plans. Last reviewed November 2026.

RapidTextAI dashboard with article generation and model tools

What the MCP Server Does

One endpoint that lets an agent create content with your RapidTextAI account, without building a separate integration for every model.

Generate articles

Create a complete article from a topic, keywords, tone, style, structure, length, language, and call to action. Results return as Markdown.

Run the agent pipeline

Use a five-step premium pipeline: draft, critique, revise, SEO optimize, and polish. Higher quality output, with a higher request cost.

Research, check and refresh

Write a sourced Pro Article with live web research, find what top pages cover that you do not, fact check claims, check originality, refresh old posts, and fill a brand profile from a website.

Discover models and usage

List available models by provider and check the authenticated user's request limit, usage, and remaining requests before generating.

Endpoint and Authentication

Use your RapidTextAI API key. Bearer auth is preferred; the query parameter is a fallback for clients that cannot set headers.

SettingValue
MCP endpointhttps://app.rapidtextai.com/openai/v1/mcp-server
ProtocolJSON-RPC 2.0 over HTTP POST
Protocol version2024-11-05
AuthenticationAuthorization: Bearer YOUR_API_KEY or ?gigsixkey=YOUR_API_KEY
Content typeapplication/json
Health checkSend a GET request to the endpoint to return server info, capabilities, tools, and auth details.
Get your key: sign in to My Account and open API Key. Keep the key server-side or in your MCP client configuration. Do not place it in public code.

Available Tools

The server implements tools/list and tools/call. The two original article tools share one input schema; the v2 tools listed below take their own arguments.

ToolWhat it doesCost
generate_article Single-pass SEO article generation. Returns a Markdown article with metadata. 1 request unit for free users, 6 for subscribers, based on the article endpoint multiplier.
generate_article_agent Draft, critique, revise, SEO optimize, and polish. Returns the final Markdown article and pipeline steps. Uses four additional refinement calls on top of the article cost.
list_models Lists models from the shared catalog, optionally filtered by provider: openai, gemini, deepseek, grok, claude, or glm. Standard API request.
get_user_limits Returns requests used, request limit, remaining requests, and whether the user is a subscriber. Standard API request.
generate_pro_article Researched, sourced article from the v2 engine: live web research, FAQ, quality checks and optional images, tweets and videos. Returns title, slug, HTML body, FAQ, sources and a quality score. 6 requests.
competitor_gap Reads the top results for a keyword and returns the topics they cover that you may be missing, with a brief. 5 requests.
refresh_content Takes a url, html or text and finds outdated claims, with sourced replacements. 6 requests.
fact_check Checks each claim in your content against a quoted web source and gives a verdict of supported, partly true, contradicted or unverified, with the quoted evidence. 6 requests.
check_originality Samples distinctive sentences and searches the web for exact matches. A spot check, not a full plagiarism scan. 3 requests.
brand_from_url Reads a website and returns a draft brand profile (voice, audience, facts) for you to review. 2 requests.
list_brands Lists your saved brand profiles so an agent can pass a brand_id to generate_pro_article. Free.
get_dashboard Usage summary for the last 7, 30 or 90 days (days), including articles written and requests used. Free.

Pro Article input

Only topic is required. Research, FAQ and sources are on by default; the server runs it as a single non-streaming call so an agent gets the finished article back.

ArgumentTypeDefaultNotes
topicstringrequiredThe topic or working title.
word_countinteger1500Clamped between 300 and 4000.
researchbooleantrueSearch the web and cite sources.
include_faqbooleantrueAdd a FAQ section.
brand_idintegernoneA brand from list_brands.
embed_tweetsbooleanfalseAlso embed relevant posts, where available.

See Pro Article and the SEO tools for what each result contains.

Article tool input

Only topic is required. The defaults keep the tool predictable for agents.

ArgumentTypeDefaultNotes
topicstringrequiredThe main topic or title.
keywordsstringemptyComma-separated SEO keywords.
modelstringglm-4.6v-flashAny model ID returned by list_models.
lengthstring2000500, 1000, 2000, 3000, or 5000.
tonestringformalformal, casual, professional, friendly, or conversational.
stylestringinformativeinformative, persuasive, narrative, descriptive, or analytical.
structurestringintroduction-body-conclusionAlso supports listicle, how-to-guide, comparison, and problem-solution.
ctastringSubscribe NowCall to action integrated into the article.
languagestringEnglishArticle language.

How to Connect an MCP Client

Add the server to a remote MCP client, or bridge the HTTP endpoint to a client that only supports stdio.

1. Add the server configuration

Client configuration keys vary. This is the common JSON shape for a remote HTTP MCP server:

{ "mcpServers": { "rapidtextai": { "type": "http", "url": "https://app.rapidtextai.com/openai/v1/mcp-server", "headers": { "Authorization": "Bearer YOUR_API_KEY" } } } }

If your client uses transport instead of type, or calls the field streamable-http, follow that client's documentation. The RapidTextAI server accepts plain HTTP POST JSON-RPC requests and returns JSON responses.

2. Initialize the session

MCP clients normally send this handshake automatically. The server responds with its protocol version, capabilities, and server info.

curl -X POST "https://app.rapidtextai.com/openai/v1/mcp-server" \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "jsonrpc": "2.0", "id": 1, "method": "initialize", "params": { "protocolVersion": "2024-11-05", "capabilities": {}, "clientInfo": { "name": "my-agent", "version": "1.0.0" } } }'

3. Discover tools

curl -X POST "https://app.rapidtextai.com/openai/v1/mcp-server" \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}'

4. Generate an article

curl -X POST "https://app.rapidtextai.com/openai/v1/mcp-server" \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "jsonrpc": "2.0", "id": 3, "method": "tools/call", "params": { "name": "generate_article", "arguments": { "topic": "How to start a blog in 2026", "keywords": "blogging, SEO, content marketing", "model": "gpt-4o-mini", "length": "2000", "tone": "professional", "language": "English" } } }'

The result is returned as MCP text content containing a JSON object with the Markdown article and metadata.

5. Use the premium agent pipeline

Change the tool name to generate_article_agent with the same arguments. The response includes the final article and the completed pipeline steps.

Resources and Prompts

The server advertises two resources and two prompt templates for clients that support them.

  • rapidtextai://models lists available models.
  • rapidtextai://limits describes the authenticated user's usage limits.
  • article-brief is a prompt template for topic and keyword briefs.
  • seo-optimize is a prompt template for optimizing existing content.

See the full API documentation for the underlying chat and article endpoints.

Usage and Limits

MCP calls count against the same API request allowance as the rest of RapidTextAI. The server checks the user's remaining requests before running an article tool, so an agent cannot start a job that will exceed the limit.

  • Free accounts have a smaller request allowance.
  • Subscriber article generation uses the article endpoint multiplier.
  • Agent mode adds four refinement calls to the article cost.
  • Unused requests do not roll over.

Compare plans to find the right allowance for your agent workload.

MCP Server FAQ

Common questions before connecting an agent.

MCP is an open protocol that lets AI applications call external tools and resources in a standard way. The RapidTextAI MCP Server exposes article generation, model discovery, and usage checks as MCP tools.

Any MCP client that supports remote HTTP JSON-RPC servers. If your client only supports stdio, use an HTTP-to-stdio bridge. Configuration key names vary by client, so check the client documentation.

The MCP endpoint returns a complete JSON-RPC response. Internally it streams from the RapidTextAI article endpoint and returns the accumulated article, so clients receive the finished result rather than live token updates.

API access is included in paid plans. MCP calls use the same API request allowance as the rest of your account. See the pricing page for current limits.

Use Bearer authentication and keep the key in your MCP client's local configuration or a server-side secret store. Do not commit it to a repository or expose it in client-side code. Rotate the key if it is exposed.

Ready to connect your
agent to RapidTextAI?

Start Free Trial
cta-area