Render Skills for AI Agents

A catalog of 21 skills that teach AI coding tools how to deploy, operate, and debug apps on Render. Each skill is a self-contained SKILL.md plus references that any compatible agent can use.

Use this repo to:

• Install Render skills into your AI tool (Claude Code, Codex, Cursor, OpenCode)
• Browse what each skill does and link straight to its docs
• Contribute new skills or improve existing ones

Compatibility

Installation

The recommended way to install Render skills is the Render CLI:

render skills install

Other useful commands:

render skills          # interactive management
render skills list     # show installed skills
render skills update   # update installed skills

Other ways to install

• Skills CLI: npx skills add render-oss/skills
• Claude Code plugin: /plugin marketplace add render-oss/skills then /plugin install render@skills
• Manual: copy any directory from skills/ into your tool’s skills path (~/.claude/skills/, ~/.codex/skills/, ~/.config/opencode/skills/, or ~/.cursor/skills/)

Prerequisites

• A Render account with the Render MCP server configured, or the Render CLI installed
• A RENDER_API_KEY environment variable
• A Git repository on GitHub, GitLab, or Bitbucket for any deploy-related skill

Skills catalog

Get started

Service types

Build and runtime

Networking and access

Data services

Operate and scale

Trying it out

Once installed, ask your agent things like:

• “Deploy my application to Render.”
• “Debug why my Render service won’t start.”
• “Is my Render service healthy?”
• “Set up a private service for this internal API.”
• “Add a cron job that runs every night.”
• “Configure custom domains for this web service.”
• “Migrate my Heroku app to Render.”

Auto-approval hooks (Claude Code)

The repo ships hook configuration that lets Claude Code auto-approve safe, read-only Render CLI operations:

• Listing services: render services list, render services -o json
• Reading logs: render logs -r, render logs --raw
• Checking workspaces: render workspace current, render workspace list

Only a single render command made of plain words is auto-approved. Commands containing quotes, pipes, redirects, &&, ;, $(...) or other shell syntax fall through to Claude Code’s normal permission checks. Malformed hook input and missing jq also produce no approval decision.

Anything that changes infrastructure still requires explicit approval, including deploys, restarts, service creation and deletion, configuration changes, and workspace changes.

Contributing

Add a new skill

1. Create skills/your-skill-name/ with a SKILL.md.

2. Add supporting files as needed: references/, assets/, and evals.json.

3. Add frontmatter at the top of SKILL.md:

   ---
   name: your-skill-name
   description: One sentence describing when an agent should use this skill.
   license: MIT
   compatibility: Prerequisites and requirements
   metadata:
     author: Render
     version: "1.0.0"
     category: deployment
   ---
   

4. Keep SKILL.md short and action-oriented. Move depth into references/.

5. Test locally with your target tool.

For reference, look at render-deploy, render-debug, render-monitor, or render-workflows.

Repository structure

skills/
├── .github/workflows/   # CI
├── hooks/               # Auto-approval hook config for Claude Code
├── scripts/             # Install and helper scripts
├── skills/              # 21 skill directories
├── README.md
└── LICENSE

Support

• Documentation:
• Issues:
• Render Support:

License

MIT License. See LICENSE.