Get Started

    Using Render with Coding Agents

    Deploy and manage apps using LLM-powered tools.

    Render integrates with popular coding agents to help you manage your infrastructure, diagnose issues, and learn about the platform.

    Agent skills

    Render's official skills enable agents to deploy, debug, and monitor your apps. Use skills with popular tools like Codex, Claude Code, and Cursor.

    View skills on GitHub

    Install

    Quick-install

    Run the following script to automatically detect popular supported tools (Claude Code, Cursor, etc.) on your system and install Render's official skills for each:

    The output looks like this:

    Running the quick-install script
    Codex installation

    Render provides a curated skill in the Codex skill installer. Run the following in Codex:

    Manual installation

    1. Clone the Render Skills GitHub repository.

    2. From the repo's skills/ directory, copy each skill directory into your tool's skills directory.

      Default skills directories for popular tools are listed below:

      ToolSkills Directory

      Claude Code

      ~/.claude/skills/<skill-name>/

      Codex

      ~/.codex/skills/<skill-name>/

      Cursor

      ~/.cursor/skills/<skill-name>/

      OpenCode

      ~/.config/opencode/skills/<skill-name>/

    Depending on your tool, you might need to restart it after installing.

    After installing, you can invoke the skill using your tool's syntax (usually a / or $ symbol followed by the skill name). Your tool also invokes the skill automatically to fulfill relevant prompts.

    Invoking the render-deploy skill in Codex

    Supported skills

    SkillDescription

    render-deploy

    Deploy applications using IaC with Render Blueprints or directly via MCP.

    Includes automatic codebase analysis and environment variable management.

    render-debug

    Debug deployment issues using logs, metrics, and database queries.

    Detects common issues like missing environment variables, port binding errors, and resource constraints.

    render-monitor

    Monitor service health, performance metrics, logs, and resource usage in real-time.

    Example prompts

    Ask your tool to perform any of these tasks:

    Deploy my application to Render
    Debug my Render deployment
    Is my Render service healthy?

    Your tool will invoke the appropriate set of skills and guide you through the process.

    Jules integration

    Jules by Google Labs provides a managed integration with Render. Whenever you open a pull request in your service's repo, Jules can detect failures in your service's preview build and automatically push fixes to address them.

    Prerequisites

    • Your Render service's repo must be hosted on GitHub.
    • Jules must have access to your service's repo.
    • Pull request previews must be enabled for your service.
      • These are the preview builds that Jules uses to detect and address issues.

    Setup

    1. Go to dashboard.render.com/jules.

      This opens the API Keys section of your user settings with a Jules-specific section:

      Jules API Keys section

    2. Next to the Jules by Google Labs section, click + Create API Key.

      A creation dialog appears.

    3. Review and accept the terms for Render's Jules integration, then click Create API Key.

    4. Copy the created API key to your clipboard.

    5. Open your Jules integrations page:

      Jules integrations page

    6. Under the Render integration, paste the API key you copied and submit it.

    You're all set! Whenever a pull request preview fails for your repo, Jules will automatically analyze its logs to identify the root cause and push a fix to address it.

    You can also enable MCP features to use Render's official MCP server with the Jules agent:

    Enabling MCP features in Jules

    You can disconnect the integration at any time by deleting the API key from the Jules integrations page.

    Render MCP server

    Connect to Render's official MCP server to manage your Render infrastructure directly from apps like Cursor and Claude Code:

    The MCP server provides tools for actions such as:

    • Spinning up new services
    • Querying databases
    • Analyzing metrics and logs

    It's especially useful for helping you identify and resolve issues with service deploys.

    Get started with Render MCP

    Documentation features

    The Render documentation provides the following capabilities to improve content discoverability and parsing for LLMs:

    Articles as markdown

    Each article under render.com/docs/ is available in a simplified markdown format that's well suited for LLMs.

    Obtain an article's markdown version by doing any of the following:

    • Append .md to the end of an article's URL:

    • Include an Accept: text/markdown header in your request to an article's URL (no .md extension required).

      • Agentic tools like Claude Code often include this header in their HTTP requests by default.

    • Click Copy page in the top-right corner of an article to copy its markdown version to your clipboard (not available on smaller screen widths).

    Reading llms.txt and llms-full.txt

    The Render documentation includes llms.txt and llms-full.txt files at the following URLs:

    FileDescription

    llms.txt

    Provides a summary of the Render documentation, including links and descriptions of each article.

    llms-full.txt

    Combines most of the Render documentation into a single, simplified markdown file. Some content types are omitted for brevity.

    Docs via MCP

    This feature is experimental.

    Render might discontinue support for this documentation-specific MCP server at any time in the future.

    Render's primary MCP server does not yet provide tools for querying the Render documentation.

    To query the Render docs from LLMs, you can connect your app or agent to the following additional MCP server:

    This MCP server provides "tools" for searching and asking questions about the Render docs. It uses the same LLM-powered answer engine as the Ask AI assistant in the documentation.