About HiveAudit Readiness
HiveAudit Readiness is an MCP server for organizations assessing AI compliance across multiple jurisdictions. It returns penalty exposure, cited gaps, a suggested audit tier, and the nearest enforcement deadline; its scoring is not legal advice.
Key features
- 01Multi-jurisdiction readiness scoring
- 02Penalty exposure in EUR and USD
- 03Article-citing compliance gaps
- 04Suggested audit tier and deadline
- 05Published tier pricing lookup
Use cases
- 01Assess readiness across applicable AI regulations
- 02Identify cited compliance gaps and potential penalties
- 03Choose an audit tier and review its price
MCP support
Exposes readiness scoring and audit-tier pricing as tools to MCP clients, backed by the HiveAudit assessment endpoint.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://hive-mcp-audit-readiness.onrender.com/mcpclaude mcp add --transport http srotzin-hive-mcp-audit-readiness https://hive-mcp-audit-readiness.onrender.com/mcp{"mcpServers":{"srotzin-hive-mcp-audit-readiness":{"type":"http","url":"https://hive-mcp-audit-readiness.onrender.com/mcp"}}}Open in Cursor{"servers":{"srotzin-hive-mcp-audit-readiness":{"type":"http","url":"https://hive-mcp-audit-readiness.onrender.com/mcp"}}}Open in VS Code{"mcpServers":{"srotzin-hive-mcp-audit-readiness":{"type":"http","url":"https://hive-mcp-audit-readiness.onrender.com/mcp"}}}{"mcpServers":{"srotzin-hive-mcp-audit-readiness":{"serverUrl":"https://hive-mcp-audit-readiness.onrender.com/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEhive-mcp-audit-readiness
EU AI Act · Colorado AI Act · CCPA · SB 942 · LL 144 · HIPAA — readiness scoring with sourced penalty math
https://hive-mcp-audit-readiness.onrender.com/mcp
What this is
hive-mcp-audit-readiness is a Model Context Protocol (MCP) server that wraps the live HiveAudit Readiness assessment endpoint at https://hivemorph.onrender.com/v1/audit/readiness. It computes a multi-jurisdictional AI compliance readiness score in a single call: penalty exposure in EUR and USD, Article-citing compliance gaps, recommended audit tier, and the nearest enforcement deadline.
Penalty math is sourced from the regulations directly:
-
EU AI Act — Reg 2024/1689 Art 99 (€15M floor, €35M cap, 3% / 7% turnover)
-
Colorado AI Act — CRS § 6-1-1701 ($20K per consumer violation, effective 2026-06-30)
-
California SB 942 — Cal. Bus. & Prof. § 22757 ($5K / violation / day)
-
CCPA — Cal. Civ. § 1798.155 ($7,500 per violation)
-
NYC LL 144 — § 20-871 ($1,500 per occurrence)
-
HIPAA — 45 CFR § 160.404 ($1.5M annual cap)
-
NIST AI RMF + ISO 42001 — gap analysis only, no penalty math
-
Protocol: MCP 2024-11-05 over Streamable-HTTP / JSON-RPC 2.0
-
Transport:
POST /mcp -
Discovery:
GET /.well-known/mcp.json -
Health:
GET /health -
Auth: none — free
-
Rate limit: 10 requests / IP / hour
-
Brand gold: Pantone 1245 C /
#C08D23
Tools
| Tool | Description |
|---|---|
audit_readiness_score | Compute a multi-jurisdictional readiness score: penalty exposure (EUR + USD), Article-citing compliance gaps, recommended audit tier, nearest deadline. |
audit_get_tier_pricing | Get the four published HiveAudit tier prices and brackets (STARTER $500 / STANDARD $1,500 / ENTERPRISE $2,500 / FEDERAL $7,500/yr). |
Backend endpoints
| Method | Path | Purpose |
|---|---|---|
POST | /v1/audit/readiness | Multi-jurisdictional readiness score |
Run locally
git clone https://github.com/srotzin/hive-mcp-audit-readiness.git
cd hive-mcp-audit-readiness
npm install
npm start
# server up on http://localhost:3000/mcp
curl http://localhost:3000/health
curl http://localhost:3000/.well-known/mcp.json
Connect from an MCP client
Claude Desktop, Cursor, Manus
{
"mcpServers": {
"hive-audit-readiness": {
"url": "https://hive-mcp-audit-readiness.onrender.com/mcp",
"transport": "streamable-http"
}
}
}
Coinbase AgentKit (Model Context Protocol extension)
import { AgentKit } from "@coinbase/agentkit";
import { McpClient } from "@modelcontextprotocol/sdk/client";
const audit = new McpClient();
await audit.connect({
url: "https://hive-mcp-audit-readiness.onrender.com/mcp",
transport: "streamable-http"
});
const agentKit = await AgentKit.from({
cdpApiKeyId: process.env.CDP_API_KEY_ID,
cdpApiKeySecret: process.env.CDP_API_KEY_SECRET,
mcpClients: [audit]
});
GOAT SDK
GOAT SDK consumes any MCP server via its @goat-sdk/plugin-mcp adapter — point at https://hive-mcp-audit-readiness.onrender.com/mcp.
Pulse MCP
Auto-crawled from the Official MCP Registry entry io.github.srotzin/hive-mcp-audit-readiness.
Glama / MCP.so / mcpverse / mcp-hunt / mkinf
Auto-crawled from this GitHub repo.
Example call
curl -s -X POST https://hive-mcp-audit-readiness.onrender.com/mcp \
-H 'Content-Type: application/json' \
-d '{
"jsonrpc":"2.0",
"id":1,
"method":"tools/call",
"params":{
"name":"audit_readiness_score",
"arguments":{
"organization_country":"DE",
"jurisdictions":["EU"],
"data_volume_records":500000,
"agent_count":10,
"monthly_inference_calls":5000000,
"sectors":["finance"],
"frameworks":["eu_ai_act","gdpr"],
"company":"Example GmbH"
}
}
}'
Returns penalty exposure (€15M EU AI Act Art 99), specific gaps (Articles 12, 13), and the recommended ENTERPRISE tier ($2,500) with a 30-day Professional trial CTA.
Disclaimers
- This is an automated readiness score, not legal advice. Engage qualified counsel for binding compliance opinions.
- Penalty math reflects statutory ceilings as of 2026-04-30. Statutes change — verify the regulation source if relying on a specific number.
- The HiveAudit Professional trial is a 30-day evaluation. The four tier prices are one-time bundles or annual subscriptions; see thehiveryiq.com/pricing.
License
MIT — see LICENSE.
Brand gold: Pantone 1245 C / #C08D23. Treasury (EIP-55): 0x15184Bf50B3d3F52b60434f8942b7D52F2eB436E (Base mainnet).
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.