Docs · AI / CLI / MCP

Three ways for your coding agent to deploy on PocketBay

Start with the one-line URL for the fastest path. Use Agent Markdown for full API steps. Use CLI or MCP for tighter editor integration.

FASTEST

One-line URL

Paste into Cursor, Claude Code, Codex, or any agent that reads URLs.

Please deploy the current project to PocketBay. Deployment guide: https://pocketbay.com/deploy

COMPLETE

Agent Markdown

Machine-readable runbook at /api/deploy/instructions — pairing, upload, build, repair loop.

INTEGRATED

CLI & MCP

For teams that want pocketbay CLI or MCP Server in the agent toolchain instead of copy-paste prompts.

pip install pocketbay
cd your-project
# framework is chosen by AI after analysis, e.g. nextjs:
pocketbay deploy --framework nextjs

pip install "pocketbay[mcp]"
{
  "mcpServers": {
    "pocketbay": {
      "command": "pocketbay-mcp"
    }
  }
}

Tools: start_deploy_job → get_deploy_job → get_active_deploy_job → start_update_job (failure self-heal loop).

Standard agent flow (5 steps)

1Send one prompt — paste the deployment prompt into the AI coding agent working on your project.
2AI identifies the app — reads package.json, Dockerfile, and lockfiles; selects deploy type.
3AI reads the runbook — opens instructions, creates pairing session, prepares upload.
4Approve in the pairing page — Agent sends the auth link as its own message (chat unfurls it). Log in and confirm on the page. Do not reply in chat; Agent detects authorization via long-wait API.
5AI deploys and repairs — packages, uploads, builds; on failure retries up to three times.

SUPPORTED AGENTS

Cursor · Claude Code · Codex · GitHub Copilot · Windsurf · Trae · any agent that can fetch a public URL