PromptGuard
PromptGuard Security is a MCP server that scans prompts for injection, redacts PII, and audits LLM SDK usage from any MCP client.
About
PromptGuard Security MCP wraps PromptGuard’s API as a local stdio Model Context Protocol server so developers can gate risky text before it hits models or users. developers shipping AI features in Claude Code, Cursor, or custom agents get a consistent place to scan for prompt injection patterns, strip or mask PII, and review how LLM SDKs are invoked—reducing accidental data leaks and jailbreak paths without building a bespoke security pipeline. The registry lists three install paths—Python promptguard-mcp-server, npm @promptguard/mcp-server, and a Docker image—plus optional PROMPTGUARD_API_URL for self-hosted backends. This fits the ship phase security lane alongside code review and secrets hygiene. It complements agent skills that generate code; it does not replace threat modeling or penetration testing of your full stack.
- stdio MCP server for injection detection on prompts
- PII redaction suitable for logging and agent transcripts
- LLM SDK usage auditing from any MCP-compatible client
- Shipped on PyPI, npm (@promptguard/mcp-server), and OCI Docker image 1.0.0
- PROMPTGUARD_API_KEY required (pg_sk_test_ or pg_sk_prod_ prefixes)
PromptGuard by the numbers
- Data as of Jul 7, 2026 (Skillselion catalog sync)
claude mcp add --env PROMPTGUARD_API_KEY=YOUR_PROMPTGUARD_API_KEY --env PROMPTGUARD_API_URL=YOUR_PROMPTGUARD_API_URL promptguard-mcp-server -- uvx promptguard-mcp-serverAdd your badge
Show developers this MCP server is listed on Skillselion. Paste this into your README.
| Package | promptguard-mcp-server |
|---|---|
| Transport | STDIO |
| Auth | Required |
What it does
Harden agent and LLM apps by scanning prompts for injection, redacting PII, and auditing SDK usage from any MCP client.
Who is it for?
Best when you're adding LLM endpoints, RAG, or tool-using agents and want PromptGuard checks inside your existing MCP workflow.
Skip if: Skip if you need full WAF, SOC2 evidence packs, or security review with no API key budget or no LLM surface in the product.
What you get
With PromptGuard wired via stdio MCP, your agent can call security checks so risky prompts and sensitive data are caught before production traffic.
- Runnable PromptGuard MCP tools for injection scan, PII redaction, and SDK audit
- Configurable API URL for self-hosted PromptGuard backends
- Repeatable security checks callable from agent sessions
By the numbers
- Version 1.0.0
- 3 distribution channels: PyPI, npm, OCI Docker
- stdio transport only in registry
Recommended MCP Servers
1Claw VaultHSM-backed vault secrets for AI agents (JIT fetch) plus prompt-injection and threat scanning.2
1passwordMCP server for 1Password service accounts — tools and resources for vaults and credentials18
402sentinel McpAssess an x402 counterparty's risk BEFORE paying: allow/review/block, scored on-chain (Base).1
A2adependencyinspector McpRemote MCP for A2A dependency inspector MCP, structured receipts, audit logs, and reviewer-ready evi
A2a Governance Bridge McpA2A Governance Bridge MCP server.A2aidentitytoll McpRemote MCP for A2A caller identity, scope policy, verdict receipts, and audit history.How it compares
LLM prompt security MCP, not a general dependency vulnerability scanner or code review skill.
FAQ
Who is PromptGuard Security MCP for?
Developers shipping AI-powered apps and agents who need injection scanning, PII handling, and SDK audits without leaving their MCP setup.
When should I use PromptGuard Security MCP?
Use it in ship (and operate) whenever user-generated or retrieved text flows into models, tools, or logs in production-bound code.
How do I add PromptGuard Security to my agent?
Install promptguard-mcp-server from PyPI, @promptguard/mcp-server from npm, or the OCI image, set PROMPTGUARD_API_KEY, and register the stdio server in your MCP client config.