# Scorable

> Scorable (formerly Root Signals) is an AI evaluation platform. It helps engineering teams measure, test, and improve LLM-powered applications using custom judges, automated experiments, and production monitoring.

## Product

- [Home](https://scorable.ai): Overview of Scorable's AI evaluation capabilities.
- [Pricing](https://scorable.ai/pricing): Free tier and paid plans for teams.
- [Book a Demo](https://scorable.ai/book-a-demo): Schedule a walkthrough with the team.
- [Use Cases](https://scorable.ai/use-cases): How teams evaluate chatbots, RAG pipelines, and AI agents with Scorable.
- [Quick Start](https://scorable.ai/quick-start): Create an LLM-as-a-judge evaluator and score output in minutes.
- [LLM-as-a-Judge](https://scorable.ai/c/llm-as-a-judge): Reliable AI with comprehensive LLM-as-a-judge evaluation.
- [LLM Guardrails](https://scorable.ai/c/guardrails): Real-time prevention of AI failures in production agents.
- [AI Agent Evaluation](https://scorable.ai/c/ai-agent-evaluation): Score every agent response for accuracy, safety, and compliance.

- [Blog](https://scorable.ai/blog): 100+ articles on LLM evaluation, guardrails, RAG metrics, and AI reliability.
- [Events and Webinars](https://scorable.ai/events-and-webinars): Talks and webinars on AI evaluation.

## Documentation

- [Docs](https://docs.scorable.ai/llms.txt): Full documentation in LLM-friendly format.
- [Full content](https://scorable.ai/llms-full.txt): Site content index - marketing, legal, and campaign pages plus all blog posts with summaries.

## API

- [API Reference](https://docs.scorable.ai): REST API and SDK documentation.
- [MCP Server](https://docs.scorable.ai/mcp-server.md): Hosted remote MCP server at https://api.scorable.ai/mcp - 14 tools for creating and running evaluations.
- [Authentication](https://scorable.ai/auth.md): How agents get and use API keys (bearer auth, free demo keys).

## MCP server

Scorable runs a hosted remote MCP server at `https://api.scorable.ai/mcp`. Connecting an MCP
client to it gives an agent 14 tools to find, run, and author LLM evaluations: `list_judges`,
`get_judge`, `run_judge`, `list_evaluators`, `get_evaluator`, `run_evaluator`,
`generate_judge`, `create_judge`, `update_judge`, `create_evaluator`, `update_evaluator`,
`list_execution_logs`, `get_execution_log`, and `list_projects`. Nothing is installed and no
local process runs; authenticate with a Scorable API key.

Claude Code:

```bash
claude mcp add --transport http scorable https://api.scorable.ai/mcp \
  --header "Authorization: Bearer $SCORABLE_API_KEY"
```

Codex, in `~/.codex/config.toml`:

```toml
[mcp_servers.scorable]
url = "https://api.scorable.ai/mcp"
bearer_token_env_var = "SCORABLE_API_KEY"
```

Cursor, VS Code, and other streamable-HTTP clients take the URL with an
`Authorization: Bearer <key>` header. `run_judge` scores a single request/response pair or a
whole conversation, and returns a score between 0 and 1 with a written justification for each
evaluator. Get a free key with
`curl -s -X POST https://api.scorable.ai/create-demo-user/ | jq -r .api_key`.

- [MCP Server docs](https://docs.scorable.ai/integrations/mcp-server): Install instructions, the full tool list, and troubleshooting.

## Agentic integration instructions

https://scorable.ai/SKILL.md