Skip to main content

Documentation Index

Fetch the complete documentation index at: https://withcoral.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

Coral cover Agents are able to make fewer, more precise tool calls with Coral than they do with data source MCP servers, CLI tools or API wrappers. For agent read tasks, SQL has a structural advantage when query complexity exceeds what a single API call can answer. It avoids pagination through large results, has cleaner, tabular responses, can bring back specific columns, and can correlate across sources more efficiently. Coral supports a number of popular data sources bundled in, and you can easily extend it to accommodate others by writing your own source specs. You can run SQL from the CLI or through MCP. And everything is local; your data, credentials and usage history never leave your machine.

​
Get started

1

Install Coral

Install the Coral CLI to get started.
brew install withcoral/tap/coral
See all installation options
2

Add your sources

Connect GitHub, Slack, Datadog, and other bundled sources to your workspace. For example:
coral source add --interactive github
Go to the quickstart
3

Start querying

Write SQL directly or let your AI agent query on your behalf.
coral sql "SELECT name, stargazers_count FROM github.org_repos WHERE org = 'withcoral' ORDER BY stargazers_count DESC"
Or use Coral over MCP

​
Why Coral

Most agent workflows access company data one tool at a time. That works, but it tends to create:
  • too many tool calls
  • repeated auth, pagination, and retry logic
  • poor cross-source reasoning
  • high token traffic
  • brittle glue code and prompts
Coral gives agents one query interface instead:
  • query multiple live sources through SQL
  • keep workflows inspectable and scriptable
  • expose the same runtime over MCP
  • answer cross-source questions without stitching tools together by hand

​
Benchmark

We benchmarked Coral against direct provider MCPs (Datadog, Sentry, Linear, Slack, and GitHub) for a diverse set of 82 real-world AI tasks using Claude Opus 4.6. Key findings:
  1. Widespread impact on performance. Across all tasks, Claude was 20% more accurate and 2x more cost efficient using Coral than using direct provider MCPs. With Coral, Claude also had 42% lower latency.
  2. Highest impact on coding agent tasks. Across the more complex tasks that typify coding agent workloads (multi-hop, higher post-processing), Claude was 31% more accurate and 3.4x more cost efficient with Coral.
  3. More neutral impact on simpler tasks. For simpler AI tasks, such as raw fact retrieval from knowledge bases, the results were closer, with Claude 6% more accurate and 2% more cost efficient with Coral.
Full benchmark report.

​
How Coral works

Coral sits between your agents and your data sources: your agents write SQL, and Coral translates it into API calls or file reads, then returns a single query result. You can ask your agents complex questions about your data: coral sql demo Or run SQL queries yourself: coral sql demo A source spec is a YAML file that defines how to reach an API or local dataset and which tables/columns it exposes. A source is a data source Coral can query, created from a source spec plus your configured credentials and variables. When you run coral source add github, Coral installs the github source. At query time, Coral loads that source as the github SQL schema, so tables like github.issues and github.pulls are queryable. Start with bundled sources or write your own. During source add, Coral collects each declared variable and secret (tokens, workspace IDs, file paths, etc.) from environment variables of the same name, or prompts for them interactively when you pass --interactive. These values are stored locally in Coral state, with secrets kept separately from non-secret config, and used only at query time. Because each source appears as SQL tables, you can JOIN across sources in one statement (for example github.issues with linear.attachments), and Coral executes that locally on your machine. For the full internals, crates, gRPC transport, DataFusion integration, see the architecture page.

Use Coral over MCP

Set up MCP for Claude Code, Cursor, and other agents

Write a custom source

Connect any API or dataset to Coral

Bundled sources

GitHub, Slack, Stripe, and more

Source spec reference

Full YAML field reference for source specs