Rail Score
Connects Claude, Cursor, or any MCP client to Responsible AI Labs' scoring and guardrail API. You get nine tools: evaluate content across eight safety dimensions with optional policy enforcement, detect prompt injection in untrusted input, gate tool calls before execution, scan tool outputs for PII and return redacted text, and check compliance against GDPR, CCPA, HIPAA, and India's DPDP Act. Reach for this when you're building agentic workflows that handle user input or need to meet regulatory requirements. The server is stateless HTTP, runs hosted at their domain, and forwards requests to their existing REST API using your RAIL API key. No SDK required, just add the URL and your key to your MCP config.
- tools
- 9
- transport
- HTTP
- last updated
- Jun 12, 2026
Tools
9 toolsPublic tool metadata for what this MCP can expose to an agent.
rail_evaluateScore content across the 8 RAIL dimensions (optional `policy` enforcement)
Score content across the 8 RAIL dimensions (optional `policy` enforcement)
No parameter schema in public metadata yet.
rail_check_complianceCheck against gdpr, ccpa, hipaa, eu_ai_act, india_dpdp, india_ai_gov
Check against gdpr, ccpa, hipaa, eu_ai_act, india_dpdp, india_ai_gov
No parameter schema in public metadata yet.
rail_detect_injectionDetect prompt injection in untrusted text
Detect prompt injection in untrusted text
No parameter schema in public metadata yet.
rail_evaluate_tool_callAllow/warn/block a tool call before it runs
Allow/warn/block a tool call before it runs
No parameter schema in public metadata yet.
rail_scan_tool_resultScan a tool's output for PII + injection, return redacted text
Scan a tool's output for PII + injection, return redacted text
No parameter schema in public metadata yet.
rail_safe_regenerateIteratively regenerate content until it passes (slow)
Iteratively regenerate content until it passes (slow)
No parameter schema in public metadata yet.
rail_dpdp_scanScan for Indian personal data under the DPDP Act 2023
Scan for Indian personal data under the DPDP Act 2023
No parameter schema in public metadata yet.
rail_dpdp_gateReal-time DPDP processing gate (allow/block/require_action)
Real-time DPDP processing gate (allow/block/require_action)
No parameter schema in public metadata yet.
rail_dpdp_complianceDPDP workflow: emit, require, evidence, session, timers
DPDP workflow: emit, require, evidence, session, timers
No parameter schema in public metadata yet.
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageReadme
RAIL Score MCP Server
Add a responsible-AI safety layer to any agent in one URL.
A remote, hosted Model Context Protocol server that exposes RAIL Score's evaluation, agent-guardrail, and India DPDP compliance capabilities to any MCP client — Claude, ChatGPT, Cursor, Copilot, Replit Agent, LangGraph, CrewAI, or a custom stack — with zero SDK integration.
https://mcp.responsibleailabs.ai/mcp
The server is a thin, hardened gateway in front of the existing REST API at
api.responsibleailabs.ai/railscore/v1/. It reimplements no scoring logic: it
validates the caller, shapes requests and responses for agent ergonomics, and
forwards to the engine. Credits, tenancy, and rate limits are identical via MCP
and REST.
Quickstart
You need a RAIL API key (rail_...) from the dashboard.
Claude Code
claude mcp add --transport http rail https://mcp.responsibleailabs.ai/mcp \
--header "Authorization: Bearer ${RAIL_API_KEY}"
Cursor / Windsurf (.cursor/mcp.json)
Details
- Repository
- github.com/responsible-ai-labs/rail-score-mcp
- Website
- responsibleailabs.ai
- Transport
- HTTP
- Registry
- MCP registry, active