Log inSubmit server
Shrike Security logo

Shrike Security

Scans prompts, responses, tool calls, queries, and agent messages for threats before execution.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Shrike Security

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "shrike-security": {
      "command": "npx",
      "args": [
        "-y",
        "shrike-mcp"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Automated check passed— started and listed 15 tools correctly (23d ago).

Overview

Shrike Security MCP server evaluates prompts, responses, tool calls, queries, file writes, commands, searches, and agent messages before they are processed or executed. It runs as an npm-based stdio MCP server through npx and returns policy decisions that allow, flag, or block actions. Anonymous use provides pattern-based detection without an API key, while registered tiers add higher limits and, on Pro or Enterprise, deeper semantic and session analysis. Reach for it when an MCP-compatible agent needs pre-execution security checks across multiple action types.

Use cases

  • Scan prompts before sending them to a model
  • Check SQL and shell commands before execution
  • Inspect file writes for traversal and secret exposure
  • Review agent-to-agent messages and AgentCards
  • Require approval for flagged agent actions

Key features

  • Pre-execution scanning for prompts, queries, commands, and writes
  • Response, A2A message, AgentCard, and MCP schema checks
  • Human approval workflow for flagged actions
  • Nine-layer detection pipeline by subscription tier
  • Session correlation and task-scope controls
  • Anonymous operation without an API key

Tools15

What Shrike Security lets your agent do. Expand a tool for its parameters.

Verified live~9.9k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 1d ago

Recent health check succeeded.

19/20 checks healthy over the last 90d

What it can do

From the 15 tools this server returned when we ran it; 14 declare their own safety hints.

Runs commands · 4Handles credentials · 7Makes network requests · 13Read-only tools · 11
Unusual tool instructions. 1 tool description contain text aimed at the AI model rather than at you (reaches for local secrets). Read the tool descriptions before connecting.
Network at startup4 connections
Files written at startupNone
Install scriptsprepare
npm provenanceNot published
Package age247 days
Maintainers1

Observed 2026-10-16 in an isolated sandbox. A description of reach, not a security audit.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 5d ago via OSV.dev · shrike-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars2Stargazers on the source repository.
npm downloads787Package downloads in the last 30 days.
Last commit1mo agoMost recent push to the default branch.
Availability100%Our rolling endpoint + install checks that succeeded.
Install checkPassedOur sandbox started it and listed its tools.
Tools exposed15Callable tools this server registers over MCP.
Directory activity1 viewsConfig copies, upvotes, and views on AllMCPs.

How Shrike Security works

What Shrike Security MCP server does

Shrike Security MCP server adds a policy checkpoint between an AI agent and the actions it wants to take. It can evaluate prompts before model processing, responses before delivery, and several classes of tool activity before execution. Decisions can allow an action, flag it for human approval, or block it.

The server covers SQL queries, file writes, shell commands, web searches, agent-to-agent messages, A2A AgentCard metadata, and MCP tool schemas. These checks target threats such as prompt injection, SQL injection, path traversal, command injection, data leakage, PII exposure, tool poisoning, and capability spoofing. Session and approval tools support longer-running workflows, including held actions and multi-turn risk tracking.

How it works

The Shrike Security MCP server uses a scan-sandwich flow: inbound prompts are checked before processing, outputs are checked before reaching the user, and tool-specific requests are checked before execution. A2A traffic can be scanned in both directions, while AgentCards and MCP schemas can be inspected before an agent trusts them.

Scans pass through a layered detection pipeline. Lower layers use pattern and content checks, including regex detection, Unicode and invisible-character checks, encoding-bypass detection, token-obfuscation detection, and embedding-based similarity analysis. Pro and Enterprise access adds visual analysis, LLM-based semantic analysis, response intelligence, and multi-turn session correlation. A cascade optimizer can stop at an earlier layer when detection confidence is sufficient.

Flagged actions can be handled through the approval tools. Session tools expose or reset correlation state, and outcome reporting records whether an allowed action was executed, failed, or skipped.

Setup and configuration

Install and run the server from npm with Node.js 18 or later.

The server runs without an API key. Anonymous requests receive L1-L5 pattern-based detection. Setting SHRIKE_API_KEY is optional and enables account-based access, including dashboard features, scan history, higher limits, and tier-dependent detection depth. Community accounts retain L1-L5 coverage, while Pro and Enterprise accounts provide the full L1-L9 pipeline.

There is no supported pip install shrike-mcp installation. Python applications should use the separately documented shrike-guard SDK instead.

Tools and capabilities

The server exposes 15 tools grouped into four functional areas:

  • Security scanners for prompts, responses, SQL, file writes, commands, web searches, A2A messages, AgentCards, and MCP schemas.
  • Approval handling for polling and submitting decisions on flagged actions.
  • Session management for reading status and resetting correlation state.
  • Scope and outcome reporting for declaring task boundaries, reporting bypasses, and recording action results.

All tools are available across the documented tiers; the tier controls scan depth, rate limits, volume, dashboard access, and session correlation.

Limitations and notes

Anonymous operation does not provide the full pipeline: L6-L9 capabilities require Pro or Enterprise access, and Community remains limited to L1-L5. Rate limits and monthly scan allowances vary by tier. The MCP package is distributed through npm and requires Node.js; the similarly named PyPI package is not the documented installation path. Account-based features require registration, and the README identifies an API key as optional rather than necessary for basic operation.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Shrike Security

Shrike Security is an MCP server that connects AI agents to Shrike’s security scanning and governance pipeline. Its main tools scan prompts, responses, SQL queries, file writes, commands, searches, agent messages, AgentCards, and MCP schemas, then allow, flag, or block actions before they proceed.

Install it through npm by configuring your MCP client to run `npx -y shrike-mcp`. Node.js 18 or later is required. The documented setup uses an MCP server entry with the `command` set to `npx` and the package name supplied in `args`.

No, an API key is not required for basic operation. Anonymous use provides L1-L5 pattern-based detection. Set `SHRIKE_API_KEY` for account-based features such as dashboard access, scan history, higher limits, and tier-dependent pipeline access.

Configure Cursor to launch the npm package with `npx -y shrike-mcp`; the same server configuration pattern is also documented for Claude Desktop, Windsurf, and Cline.

It scans prompts, model responses, SQL queries, file writes, CLI commands, web searches, A2A messages, AgentCards, and MCP schemas. Additional tools manage approvals, session state, declared scope, bypass reports, and action outcomes.

Related MCP servers

All in Databases All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewShrike Security AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/shrike-security?style=directory)](https://allmcps.com/mcp/shrike-security)
HTML Embed
<a href="https://allmcps.com/mcp/shrike-security"><img src="https://allmcps.com/api/badge/shrike-security?style=directory" alt="Shrike Security on AllMCPs" /></a>