withcoral/coral

One SQL interface over APIs, files, and live sources — built for agents.

What it solves

Coral provides a single, SQL-based query interface for AI agents to access data from various APIs and local files. It eliminates the need for agents to make multiple, complex tool calls, handle pagination, or manually stitch together results from different sources, which often leads to high token usage, increased latency, and brittle prompts.

How it works

Coral acts as a local read layer between the agent and data sources. Agents write SQL queries, which Coral translates into API calls or file reads. It then returns a tabular result set to the agent.

Key components include:

  • Sources: Defined by YAML specs that declare how to reach an API or dataset and the tables/columns it exposes.
  • Joins: Because data is presented as SQL tables, Coral can perform local joins across different sources (e.g., joining GitHub issues with Linear issues) in a single statement.
  • MCP Server: Coral includes a built-in Model Context Protocol (MCP) server, allowing agents like Claude Code, Cursor, or VS Code to interact with it as a read-only SQL database.
  • Local Execution: All data, credentials, and usage history remain on the user's machine.

Who it’s for

Developers and AI agent users who want to integrate their company data (from tools like GitHub, Slack, Datadog, and Sentry) into their AI workflows without writing custom glue code or per-source API wrappers.

Highlights

  • Unified SQL Interface: Query multiple live sources through a single language.
  • Performance Gains: Benchmarks show increased accuracy, lower latency, and higher cost efficiency compared to direct provider MCPs.
  • Local-First Privacy: Credentials and data never leave the local machine.
  • MCP Integration: Native support for the Model Context Protocol for easy connection to modern AI agents.
  • Extensible: Supports bundled sources, community source specs, and custom user-defined source specs.

Related

  • Project
    1mcp-app/agent

    A unified runtime for MCP servers that aggregates multiple servers into one process and provides a progressive discovery workflow to prevent agent context sprawl.

  • Project
    skalesapp/skales

    Skales is a closed‑source, locally‑run AI desktop and mobile agent that can chat, automate multi‑step goals, edit code, generate design/media assets, manage calendars, and retain memory—all without sending data to the cloud. It installs with a single click on Windows/macOS/Linux/Android/iOS and supports >15 LLM providers or fully offline models.

  • Project
    FerroxLabs/wayland

    A local-first desktop AI agent command center that orchestrates multiple AI CLIs and agents from one interface with shared persistent memory.

  • Project
    openyak/openyak

    A universal interface for AI agents that allows users to maintain a single persistent chat while switching between different agent runtimes like Claude Code and Codex.

  • Project
    builderz-labs/marketing-dashboard

    A local-first marketing operations dashboard that consolidates CRM, outreach, and AI agent activity into a single self-hosted interface for human-in-the-loop management.