1 Copy the MCP server URL
It's right at the top of this page.
AI Detector Bypass,
Inside Your Agent
Bring StealthGPT’s AI humanizer into Codex, Claude and your content workflows. Rewrite drafts for natural writing and detector bypass, or start a research-to-article run inside the tools your team already uses.
Connect your StealthGPT account with available subscription usage or API credits. If you have both, API credits are the default; choose subscription billing in your usage settings when eligible.
MCP ServerOne account, connected workflows
https://www.stealthgpt.ai/api/mcp/mcp
- Super, Standard and Lite humanization
- Research-to-draft content workflows
- Async results for agents and automations
Claude.ai as an example
Connect your humanization workflow
Choose your client, connect the remote server and bring StealthGPT into the conversation. Use OAuth where supported or the documented bearer-token configuration.
2 Add the connector
Follow your client’s setup guide: add the remote server URL, then connect through OAuth or configure your API token as instructed.
3 Ask your agent to humanize
Ask it to call humanize_text, save the runId and retrieve the completed result with get_run_status. Review the returned draft before delivery.
Use it inside your favorite LLM
Connect through the documented client setup. The Codex guide records a completed OAuth humanization check; the ChatGPT guide explains its custom MCP setup.
How to Connect
Choose a client and follow its authentication and setup steps.
Claude.ai & Claude Desktop
- 1.
Go to Customize section
In the sidebar, click Customize, then Connectors. (On the web you can also open claude.ai/customize/connectors directly.)
- 2.
Click the + button, then Add custom connector
In the Connectors panel, open the + menu and choose Add custom connector.
No subscription? Use MCP with an API plan.
API plans start at $19/month and include monthly words for MCP, the API, n8n, and Zapier. Extra usage is pay-as-you-go.
API plans
Monthly words, pay as you go beyond
$19per month and up
- 100,000 to 600,000 words per month (input + output)
- Writer, Humanizer & Blog Writer endpoints
- Humanization overage per 1K input + output words: Standard/Lite $0.20; Super $0.50
- Support through our support page
Super
Highest bypass rate
$0.50per 1K extra words
- Flagship model for detector bypass
- Built to bypass Pangram v4
- Available on every API plan
Enterprise
Volume discounts available
Custom
- Dedicated account manager
- Priority email & phone support
- Custom SLA agreements
- Bulk word pricing
- White-glove onboarding
- Tailored to your workflow
MCP can use eligible subscription allowances or API credits. If your account has both, check the MCP billing selection in usage settings.
Features
Call StealthGPT From Anywhere
Connect StealthGPT as a remote MCP tool so your agent can send a draft for humanization and collect the finished text in the same conversation. Use humanize_text for an existing draft, generate_content for a content brief, and get_run_status for progress and results.
Drop It Into n8n
Start and track content-generation or humanization jobs with the verified StealthGPT n8n node. Published version 0.1.2 exposes Heavy, which maps to Standard, and Lite for humanization. Use HTTP Request nodes for explicit Super selection or bulk API jobs. Store your token in an n8n credential and route completed drafts to review.
Stealth Agent: One Request, Full Pipeline
Start Stealth Agent from a brief and select an Academic, SEO or Social preset. The pipeline researches, drafts and humanizes, with optional fact checking and images. Retrieve the resulting document for your editorial review, or use humanize_text to add just the humanization step to a pipeline you already run.
Built for Volume
Start an async job, save its run ID and collect the result when it finishes. Your REST integration can use polling, signed callbacks and idempotency keys. For an existing content queue, submit up to 100 drafts through the bulk REST endpoint and collect each run for editorial review.
Why it's different
Use StealthGPT inside your connected agent or automation stack. Choose subscription usage or API credits according to your account, and keep the workflow in the tools your team already uses.
Built for humanization and detector bypass
Choose Super, Standard or Lite for your agent or application. Super quality-mode API ranked first in HumanizerEval’s October 2026 test, with an 80.3% mean strict pass rate across six external detectors.
Reliable at scale
Save the run ID, retrieve progress and collect the completed result. REST workflows can also use idempotency keys and callbacks to recover interrupted work.
Full pipeline or one step
Let Stealth Agent research, draft, fact-check, and humanize in one request, or drop humanization into the workflow you already run.
Start with one request
Start a Stealth Agent run, get a runId immediately, and collect your finished content by polling or webhook. Build on the async API with idempotency keys, webhooks, and content presets. Explore the full reference in the docs.
Stealth Agent API
POST /api/stealthify/agent/runs
api-token: YOUR_API_KEY
Content-Type: application/json
{
"preset": "seo",
"prompt": "Write about async APIs",
"webhookUrl": "https://example.com/webhooks/stealthgpt"
}
→ {
"runId": "run_abc123",
"status": "queued",
"statusUrl": "/api/stealthify/agent/runs/run_abc123"
}
// poll statusUrl with the same api-token, or receive a webhook
GET /api/stealthify/agent/runs/run_abc123
// when status is completed, read the result fieldFrequently Asked Questions
Still have more questions?
Learn more in our support page.
Yes. Eligible subscription accounts can use MCP with their plan allowance: humanize_text uses words and generate_content uses a Stealth Agent article. Connect the account you want to bill. If you also have API access, check the MCP billing rules and your usage settings before starting a larger queue.
Yes. You can connect with a StealthAPI plan. Plans start at $19/month with 100,000 billable words included. Humanization counts input plus output: Standard and Lite use a 1× multiplier and Super uses 2.5×. See API pricing for allowances and pay-as-you-go rates.
Subscription-only accounts use their plan allowance. Accounts with API access default to API credits. If you have an active subscription too, you can choose subscription billing under MCP billing in your usage settings. This choice applies to MCP; REST API calls use API credits.
MCP connects an AI client to external tools. StealthGPT’s remote server lets a compatible agent submit an existing draft for humanization or start a content workflow, then retrieve the result in the same conversation. Follow your client’s guide for authentication and tool access.
The REST API gives your software direct access to humanization, writing, detection and async jobs. Stealth Agent takes a content brief through research, drafting, optional fact checking, humanization and optional images. MCP exposes humanization and Stealth Agent tools to a connected assistant.
Yes. Connect StealthGPT through OAuth, ask Codex to call humanize_text and retrieve the result with get_run_status. Our November 7, 2026 check completed that workflow with codex-cli 0.167.0-alpha.3.2. Follow the Codex setup guide for commands and an example prompt.
Setup guides cover Codex, Claude and Cursor, with Manus instructions on this page. ChatGPT provides a custom MCP setup path, subject to account and workspace controls; it is separate from our dated Codex runtime check. n8n, WordPress and Zapier use dedicated integrations. See the client setup and verification table to choose your route.
Use humanize_text to rewrite an existing draft, generate_content to start Stealth Agent, and get_run_status to retrieve progress and results. enable_pay_as_you_go returns an account-setup link for eligible API accounts. The MCP server does not expose a standalone detection tool; use the REST detection endpoint when your workflow needs a separate detector request.
Yes. The verified n8n-nodes-stealthgpt node starts and tracks humanization and content-generation runs. Published version 0.1.2 exposes Heavy, which maps to Standard, and Lite for humanization. For explicit Super selection or bulk jobs, use HTTP Request nodes as shown in the n8n guide.
humanize_text and generate_content return a runId so your agent can retrieve progress with get_run_status. Save that ID and collect the existing run instead of submitting the draft again while it is processing. REST integrations can also use signed callbacks; the review recipes show polling, retries and result collection.
Academic, SEO and Social presets configure a research-to-draft workflow for the intended content. You can enable fact checking and image generation, then collect the resulting document for review. See the Stealth Agent guide for inputs and workflow controls.
Create a StealthGPT API credential in n8n, or use a Header Auth credential for HTTP Request nodes. Reference that credential from your workflow rather than putting the token in node text or example content. The n8n guide covers both routes.
Yes. Your agent can prepare a draft, call humanize_text and collect the completed version with get_run_status. For a brief-to-article workflow, use generate_content. To move many existing drafts through your own software, follow the async and bulk review recipes.
In HumanizerEval’s October 2026 test of 11 humanizers on the same 50 English texts, StealthGPT Super in quality mode through the API ranked first with an 80.3% mean strict pass rate across six external detectors. See the published humanizer results for test scope, settings, detector-specific rates and methodology.
The AI humanizer API comparison compares seven providers on workflows, models, pricing and data handling. The StudySim API case study shows a related XYZAI product’s Super quality-mode publishing workflow with original Search Console exports. Use the data-handling guide to plan your own integration.
Choose a setup guide, connect the server to your client and submit one draft. Select model super, standard or lite explicitly when you want a fixed workflow choice; MCP defaults to super. Retrieve the run, review the result and choose the billing source that fits your account.
Integrate StealthGPT Into Your Workflow
StealthGPT is no longer just a site you visit, it’s a tool you can call. The web app still works the way it always has, but the API, Stealth Agent, and the new MCP and n8n integrations mean the product now lives inside your agent’s reasoning chain, in your automation pipeline, and behind your webhook.