Staticbot logoStaticbot.dev
    Platform

    Staticbot MCP Server

    Give Claude Code (and any MCP-compatible AI assistant) direct control over your Staticbot infrastructure. Manage migrations, stacks, and deployments through natural language — no manual UI required.

    What is MCP?

    The Model Context Protocol (MCP) is an open standard by Anthropic that lets AI assistants like Claude securely call external tools. Instead of pasting API responses into chat, the AI calls tools directly and acts on the results — keeping your context clean and your workflow tight.

    The Staticbot MCP server runs locally alongside Claude Code and bridges natural-language instructions to the Staticbot REST API.

    36 tools across 5 categories

    Full coverage of the Staticbot workflow — from browsing templates to troubleshooting migration jobs.

    Templates

    list_templates

    List available infrastructure templates (Vite apps, Supabase stacks, etc.).

    get_template

    Get template details including configuration variables you can override.

    create_template

    Create a new template by scanning a GitHub repo. Auto-detects platforms, env vars, and builders.

    Stacks

    list_stacks

    List all infrastructure stacks in your organization.

    get_stack

    Get stack details including templates and configuration.

    create_stack

    Create a stack from a template with auto-generated, custom, or existing domain.

    Deployments

    list_deployments

    List deployments, optionally filtered by stack.

    create_deployment

    Create a deployment for a stack (does not start it yet).

    start_deployment

    Start a created deployment — provisions infrastructure in AWS.

    get_deployment

    Get deployment status. Includes required DNS records when status is WAITING.

    Migrations

    create_migration

    Create a new migration — starts the 8-phase pipeline from Discovery to Frontend Deploy.

    list_migrations

    List all Supabase migrations, optionally filtered by status.

    get_migration

    Get migration status and 8-phase breakdown.

    confirm_migration

    Approve a migration after discovery completes.

    resume_migration

    Resume a migration paused by user or awaiting action.

    pause_migration

    Pause a running migration. Current job finishes, no new jobs start.

    get_migration_jobs

    List all jobs within a migration for granular progress and troubleshooting.

    retry_migration_job

    Retry a failed migration job.

    skip_migration_job

    Skip a non-critical job that's blocking progress.

    complete_migration_job

    Complete a manual job (e.g. MANUAL_SYNC_LOVABLE) with optional functionUrl.

    get_migration_deployments

    List AWS deployments for a migration's infrastructure.

    choose_data_import_method

    Choose automated or manual data import in Phase 3.

    choose_backend_switchover

    Choose backend switchover strategy in Phase 7.

    choose_frontend_deploy

    Choose frontend deployment method in Phase 8.

    validate_function_url

    Check if a Supabase edge function URL is reachable and responding.

    list_integration_instances

    List connected integrations (Supabase, GitHub) with their type.

    parse_source_keys

    Auto-read source Supabase URL and anon key from a GitHub repo's .env file.

    list_supabase_projects

    List Supabase projects on a connected integration instance.

    lovable_sync

    Trigger Lovable sync via Chrome extension bridge to auto-deploy edge functions.

    Connected Projects (Continuous Sync)

    list_connected_projects

    List all connected projects with sync status.

    get_connected_project

    Get connected project details, sync mode, and webhook status.

    trigger_sync

    Trigger a manual sync — applies migrations, deploys edge functions, rebuilds frontend.

    list_sync_runs

    List sync run history for a project.

    get_sync_run

    Get sync run status, commit SHA, and change summary.

    get_sync_run_jobs

    Get individual jobs for a sync run to diagnose failures.

    confirm_sync_run

    Confirm a sync paused for review (destructive database migrations).

    Setup in 3 steps

    1

    Get an API key

    Log into app.staticbot.dev, go to Settings → API Keys, and create a new key.

    2

    Add to your MCP config

    Add the following to your project's .mcp.json or global ~/.claude/settings.json:

    {
      "mcpServers": {
        "staticbot": {
          "command": "npx",
          "args": ["-y", "@bitfiction/staticbot-mcp"],
          "env": {
            "STATICBOT_API_KEY": "sk-your-api-key-here"
          }
        }
      }
    }

    No install needed — npx downloads and runs the latest version automatically.

    3

    Start using it

    Restart Claude Code and run /mcp to verify the server is connected. Then just ask Claude to deploy, migrate, or manage your infrastructure.

    Example interactions

    You:

    "Deploy my Vite app to AWS on example.com"

    Claude:

    Calling list_templates → found "Vite Static Website" template.

    Calling create_stack with custom domain "example.com" → stack created.

    Calling create_deploymentstart_deployment → deployment is now IN_PROGRESS. Track progress here: [statusUrl]

    You:

    "My migration has a failing edge function. Skip it and continue."

    Claude:

    Calling get_migration_jobs → found job deploy_edge_function_hello in FAILED status. Error: "Function compilation failed."

    Calling skip_migration_job → job skipped. Dependent jobs are now proceeding.

    Why use MCP over direct API calls?

    No boilerplate

    The AI handles auth headers, JSON parsing, and error handling. You just describe what you want.

    Multi-step reasoning

    Claude can chain multiple API calls — check status, wait for readiness, then confirm — without you scripting the flow.

    Works in your codebase

    Use Staticbot tools alongside file edits and Git commands in a single Claude Code session.

    Open protocol

    MCP is supported by Claude Code, Cursor, and other editors. One server, multiple clients.

    Let your AI handle the infra

    Get an API key, install the server, and start deploying with natural language.