Deploy to Render

Deploy applications to Render using Blueprint files or direct MCP service creation.

Overview

This skill automates Render deployments for any Git-backed application. It analyzes the codebase to detect framework, runtime, build/start commands, and dependencies, then generates the appropriate deployment configuration.

What it automates:

• •Codebase analysis (framework, package manager, runtime detection)
• •Blueprint generation (render.yaml for Infrastructure-as-Code)
• •Direct service creation via MCP tools (single-service deployments)
• •Dashboard deeplink generation for one-click deployment
• •Post-deploy verification and basic troubleshooting

Time saved: ~15-30 minutes per deployment setup

Triggers

When to Run

Activate when the user wants to:

• •Deploy an application to Render
• •Create a render.yaml Blueprint file
• •Set up hosting for their project on Render
• •Create databases, cron jobs, or other Render resources
• •Troubleshoot a failed Render deployment

Manual Invocation

/render-deploy [options]

Process

Step 1: Gather Requirements

Ask the user two questions to reduce friction:

1. •Source: Git repo or prebuilt Docker image?
2. •Scope: Full provisioning (app + database + workers) or app-only?

If deploying a Docker image without a Git repo, guide the user to the Render Dashboard (MCP cannot create image-backed services).

Step 2: Choose Deployment Method

Decision rule (apply automatically unless user specifies):

• •Direct Creation when ALL true: single service, no workers/cron, no databases, simple env vars, and MCP tools are available
• •Blueprint when ANY true: multiple services, databases needed, cron/workers, reproducible IaC desired, monorepo

Default to Blueprint if unclear.

Step 3: Verify Prerequisites

Run checks in order:

# 1. Verify Git remote exists
git remote -v

# 2. Check MCP availability (for Direct Creation)
# Try: list_services()

# 3. Check Render CLI (for Blueprint validation)
render --version

# 4. Check authentication (CLI fallback)
render whoami -o json

If MCP is not configured, offer setup instructions for the user's AI tool (Claude Code, Cursor, Codex). See references/direct-creation.md for MCP setup details.

Step 4: Analyze Codebase

Detect framework, runtime, package manager, build/start commands, environment variables, and database requirements. Use the detection rules in references/codebase-analysis.md.

Step 5A: Blueprint Path

1. •

   Generate render.yaml following the specification in references/blueprint-spec.md

   • •Default to plan: free unless user specifies otherwise
   • •Mark secrets with sync: false
   • •Use appropriate service types from references/service-config.md
2. •

   Validate (if CLI available):

   bash

   render blueprints validate
   

3. •

   Commit and push:

   bash

   git add render.yaml
   git commit -m "Add Render deployment configuration"
   git push origin main
   

4. •

   Generate deeplink from the Git remote URL:

   bash

   git remote get-url origin
   

   Convert SSH to HTTPS if needed, then format: https://dashboard.render.com/blueprint/new?repo=REPOSITORY_URL

5. •

   Guide user through Dashboard: OAuth, name Blueprint, fill secrets, click Apply.

Step 5B: Direct Creation Path

1. •Create resources via MCP using the commands in references/direct-creation.md
2. •Configure environment variables after creation
3. •Services with autoDeploy deploy automatically

Step 6: Verify Deployment

Run post-deploy checks:

1. •Confirm deploy status is live
2. •Hit health endpoint and verify 200 response
3. •Scan error logs for failure signatures
4. •Verify env vars and port binding (0.0.0.0:$PORT)

See references/operations.md for verification commands and error patterns.

Verification

Success Indicators

• •Deploy status shows live
• •Health endpoint returns 200
• •No error-level log entries in recent logs
• •All required env vars are set
• •Service is accessible at its public URL

Failure Handling

Examples

Example 1: Deploy Node.js API with PostgreSQL (Blueprint)

Prompt: Deploy my Express app to Render with a database

Response: [Analyzes package.json, detects Express + pg]
Multiple services detected (web + database) - using Blueprint method.
Generated render.yaml with web service and PostgreSQL.
Validated with render blueprints validate.
Committed and pushed render.yaml.
Blueprint deeplink: https://dashboard.render.com/blueprint/new?repo=...
Click the link, fill in secrets, and click Apply to deploy.

Example 2: Quick Static Site Deploy (Direct Creation)

Prompt: Host my React app on Render

Response: [Detects single static site, MCP available]
Single service with no database - using Direct Creation.
Created static site via create_static_site MCP tool.
Build command: npm ci and npm run build
Publish path: ./build
Service deploying automatically. Checking status...
Deploy is live at https://my-frontend.onrender.com

Example 3: Dry Run Preview

Prompt: /render-deploy --dry-run

Response: [Analyzes codebase without deploying]
Framework: FastAPI (Python)
Runtime: python
Build: pip install -r requirements.txt
Start: uvicorn main:app --host 0.0.0.0 --port $PORT
Method: Blueprint (database detected)
Preview of render.yaml generated. Review and approve to proceed.

Safety

Idempotency

Blueprint generation is idempotent: running the skill again overwrites render.yaml with the same content. Direct Creation is NOT idempotent: running it again creates duplicate services.

Reversibility

• •Blueprint: Delete render.yaml and remove services from Render Dashboard
• •Direct Creation: Delete services from Render Dashboard or via MCP

Prerequisites

Before running, ensure:

• • Code is pushed to a Git remote (GitHub, GitLab, or Bitbucket)
• • You have a Render account
• • MCP tools are configured (for Direct Creation) or CLI is installed (for Blueprint validation)

Network Access

When sandboxing blocks deployment network calls, rerun with sandbox_permissions=require_escalated. The deploy requires escalated network access for outbound requests to Render.