docs: add MCP server, complete datasets table, expand links

README.md

@@ -2,7 +2,7 @@
 
 ## Mission
 
-Developer-facing AI safety infrastructure. We build scoring APIs, compliance engines, and agent safety pipelines that help teams ship AI responsibly.
+Developer-facing AI safety infrastructure. We build scoring APIs, compliance engines, and agent safety pipelines that help teams ship AI responsibly, with first-class support for India's regulatory landscape alongside global frameworks.
 
 ## What We Ship
 
@@ -14,15 +14,23 @@ Developer-facing AI safety infrastructure. We build scoring APIs, compliance eng
 
 **Safe Regeneration** automatically rewrites AI responses that score below configurable quality thresholds, with full before/after audit logging.
 
-## SDKs
+## Ways to Integrate
 
-Published on [PyPI](https://pypi.org/project/rail-score-sdk/) (Python) and [npm](https://www.npmjs.com/package/@responsible-ai-labs/rail-score) (TypeScript) with drop-in wrappers for OpenAI, Anthropic, and Gemini. Observability integrations with OpenTelemetry, Langfuse, and Datadog.
+**SDKs** published on [PyPI](https://pypi.org/project/rail-score-sdk/) (Python) and [npm](https://www.npmjs.com/package/@responsible-ai-labs/rail-score) (JavaScript/TypeScript), plus a [Drupal module](https://www.drupal.org/project/rail_score), with drop-in wrappers for OpenAI, Anthropic, and Gemini. Observability integrations with OpenTelemetry, Langfuse, and Datadog.
 
-## Open Datasets
+**MCP Server** adds the full RAIL safety layer to any [Model Context Protocol](https://modelcontextprotocol.io) client (Claude, Cursor, Copilot, Replit Agent, LangGraph, CrewAI) through a single hosted URL, with no SDK integration:
+
+```
+https://mcp.responsibleailabs.ai/mcp
+```
 
-**[RAIL-HH-10K](https://huggingface.co/datasets/responsible-ai-labs/RAIL-HH-10K)**: 10,000-example preference dataset for responsible AI alignment, scored across all 8 RAIL dimensions.
+## Open Datasets
 
-**[Indian Responsible AI Benchmark](https://huggingface.co/datasets/responsible-ai-labs/indian-responsible-ai-benchmark)**: 212 adversarial prompts across 22 India-specific safety categories, evaluated against Sarvam AI models with full RAIL dimension scores.
+| Dataset | Size | What it covers |
+|---|---|---|
+| [RAIL Guard Benchmark](https://huggingface.co/datasets/responsible-ai-labs/rail-guard-benchmark) | 1,589 examples | 1,197 content prompts across 6 domains (benign / edge / adversarial) and 392 agent tool-call scenarios across 5 domains, scored on all 8 RAIL dimensions |
+| [RAIL-HH-10K](https://huggingface.co/datasets/responsible-ai-labs/RAIL-HH-10K) | 10,000 examples | Preference dataset for responsible AI alignment (RLHF/DPO), scored across all 8 RAIL dimensions, built on Anthropic's HH data |
+| [Indian Responsible AI Benchmark](https://huggingface.co/datasets/responsible-ai-labs/indian-responsible-ai-benchmark) | 212 prompts | Adversarial prompts across 22 India-specific safety categories (caste, region, Hinglish code-switching), with full RAIL dimension scores |
 
 ## Research
 
@@ -32,5 +40,9 @@ Published on [PyPI](https://pypi.org/project/rail-score-sdk/) (Python) and [npm]
 
 - **Platform**: [responsibleailabs.ai](https://responsibleailabs.ai)
 - **Documentation**: [docs.responsibleailabs.ai](https://docs.responsibleailabs.ai)
+- **Knowledge Hub**: [responsibleailabs.ai/knowledge-hub](https://responsibleailabs.ai/knowledge-hub)
 - **GitHub**: [Responsible-AI-Labs](https://github.com/Responsible-AI-Labs)
-- **Contact**: research@responsibleailabs.ai
+- **PyPI**: [rail-score-sdk](https://pypi.org/project/rail-score-sdk/) · **npm**: [@responsible-ai-labs/rail-score](https://www.npmjs.com/package/@responsible-ai-labs/rail-score)
+- **MCP Server**: [mcp.responsibleailabs.ai/mcp](https://mcp.responsibleailabs.ai/mcp)
+- **X**: [@RAILabsIndia](https://x.com/RAILabsIndia) · **LinkedIn**: [company/rail-ai](https://www.linkedin.com/company/rail-ai/)
+- **Contact**: research@responsibleailabs.ai