Skip to content
MCP server

SafePrompt

Screen untrusted text for prompt injection and jailbreaks before it reaches your LLM

Install
Pricing
Freemium
Platforms
API, CLI
Runs
Locally or Remotely
Version
0.1.1
Listing updated
3 days ago

About SafePrompt

SafePrompt is a freemium MCP server for ai models and llm tooling. It can run locally on your machine or as a hosted remote server.

Anything an AI reads can carry instructions: a support ticket, a pasted document, a web page or another agent's message. SafePrompt checks that text first. Pattern checks block known attack shapes, and everything they do not block goes to AI analysis that reads what the message is really asking for. The reply has four fields: safe, a confidence score, the threats that fired and plain-text reasoning.

The MCP package needs a SafePrompt API key and can run over stdio from npm or at a hosted endpoint. It wraps the same validation API that the vendor sells directly, so one call can screen a user message, a retrieved document or tool output. The site publishes benchmark figures, including 4 of 949 AgentDojo attacks getting through and 92 percent of 1,000 HackAPrompt jailbreaks caught, run by the vendor in October 2026, and a median latency of 798 ms.

The free tier allows 10,000 validations a month with no card. Starter is $29 for 500,000 and Business is $99 for 1,000,000, with no overage billing. The vendor notes that blocked attacks from free traffic join a shared defence network, which paid plans can turn off.

Install

Run in a terminal

claude mcp add --transport http mcp https://mcp.safeprompt.dev/mcp

Add to Settings → Connectors → Add custom connector

https://mcp.safeprompt.dev/mcp
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "mcp": {
      "url": "https://mcp.safeprompt.dev/mcp"
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "mcp": {
      "type": "http",
      "url": "https://mcp.safeprompt.dev/mcp"
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "mcp": {
      "serverUrl": "https://mcp.safeprompt.dev/mcp"
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "mcp": {
      "type": "http",
      "url": "https://mcp.safeprompt.dev/mcp"
    }
  }
}

Configuration

  • SAFEPROMPT_API_KEYSafePrompt API key from https://dashboard.safeprompt.devRequired
  • SAFEPROMPT_PROVIDERAPI base URL (default https://api.safeprompt.dev)
  • SAFEPROMPT_USER_IPValue sent as X-User-IP for threat-intel tracking (default 203.0.113.1)

What it does

Two-stage detection
Pattern checks first, then AI analysis of messages that pass.
Structured verdict
Returns safe, a confidence score, the threats found and reasoning.
Agent input screening
Covers user messages, retrieved documents and tool output.
Published benchmarks
The vendor lists results against AgentDojo, HackAPrompt, TensorTrust and BIPIA.
Optional strict mode
A dashboard setting makes screening stricter.
LangChain wrappers
Drop-in wrappers exist for JavaScript and Python.

Good for

  • Screen user input before it reaches a chatbot
  • Check retrieved documents before an agent acts on them
  • Inspect tool output for injected instructions
  • Block jailbreak attempts in support automations
  • Add a safety gate to a LangChain app

Pricing

Freemium, from $29 a month

Free 10,000 validations/mo; Starter $29/mo for 500,000; Business $99/mo for 1,000,000

Prices change. Check safeprompt.dev before you buy.

SafePrompt FAQ

Is there a free tier?

Yes. 10,000 validations a month, no card and no time limit.

What does it need to run?

A SafePrompt API key from the dashboard, passed as SAFEPROMPT_API_KEY.

How fast is it?

The site reports a median of 798 ms across 528,248 production calls in 30 days.

What happens past my monthly quota?

Requests return HTTP 429 with the reset date; there is no overage billing.

Is SafePrompt free?

SafePrompt has a free plan, with paid plans for more usage or features. Paid plans start at $29 a month.

How do I add SafePrompt to Claude Code?

Run this command in your terminal: claude mcp add --transport http mcp https://mcp.safeprompt.dev/mcp

What are the best alternatives to SafePrompt?

The closest alternatives to SafePrompt are REA, Claude Code Ultimate Guide, Agent Roast, Midplane and Leash.

Alternatives to SafePrompt

  1. REA

    MCP server

    CLI and MCP server that lets coding agents reverse engineer apps and binaries

    Runs locally
    • Open source
    • 464K GitHub stars
    • Developer tools
  2. MCP server for searching the Claude Code Ultimate Guide from any MCP client

    Runs locally
    • Open source
    • 6.5K GitHub stars
    • AI models and LLM tooling
  3. Agent Roast

    MCP server

    Offline review that flags agent changes which weaken tests, assertions or CI

    Runs locally
    • Open source
    • 4 Hacker News points
    • Developer tools
  4. Midplane

    MCP server

    Postgres gateway for AI agents with table grants, column masking and approval for writes

    Runs locally
    • Open source
    • 5 GitHub stars
    • Databases
  5. Leash

    MCP server

    Review what your AI touched, approve risky actions and undo supported changes

    • 1 tool
    • Remote
    • Free
    • 1 GitHub star
    • Security