Deploy

End-to-end deploy of openma to your own Cloudflare account. Plan on 30–60 minutes the first time, mostly waiting for OAuth app approvals.

Prerequisites

Node 22+, pnpm 10+
A Cloudflare account on the Workers Paid plan
wrangler logged in (npx wrangler login)
A domain on your Cloudflare account
API keys for at least one LLM provider (Anthropic, OpenAI, or MiniMax)

Steps

Clone and install.

git clone https://github.com/open-ma/open-managed-agents.git
cd open-managed-agents
pnpm install

Create the Cloudflare resources.

Run these once, save the IDs they print, you’ll paste them into wrangler files.

# D1 database (shared by main + integrations)
npx wrangler d1 create openma-auth

# KV namespace
npx wrangler kv namespace create CONFIG_KV

# R2 buckets
npx wrangler r2 bucket create managed-agents-files
npx wrangler r2 bucket create managed-agents-workspace

# Vectorize index for semantic memory
npx wrangler vectorize create openma-memory \
  --dimensions=1024 --metric=cosine

Edit the wrangler files.

Open apps/main/wrangler.jsonc and replace the placeholder IDs:

{
  "d1_databases": [
    { "binding": "AUTH_DB", "database_name": "openma-auth", "database_id": "<paste yours>" }
  ],
  "kv_namespaces": [
    { "binding": "CONFIG_KV", "id": "<paste yours>" }
  ],
  "vectorize": [
    { "binding": "VECTORIZE", "index_name": "openma-memory" }
  ],
  "routes": [
    { "pattern": "app.yourdomain.com", "custom_domain": true }
  ]
}

Repeat for apps/agent/wrangler.jsonc (same D1, KV, R2 ids; agent-specific DOs are auto-created on first deploy) and apps/integrations/wrangler.jsonc (same D1).

Set secrets.

# Required
npx wrangler secret put ANTHROPIC_API_KEY        # or OPENAI/MINIMAX
npx wrangler secret put BETTER_AUTH_SECRET       # any random 32+ char string
npx wrangler secret put API_KEY                  # initial dev API key
npx wrangler secret put INTEGRATIONS_INTERNAL_SECRET   # shared between main + integrations
npx wrangler secret put MCP_SIGNING_KEY          # for outbound MCP token signing

# Per worker — repeat for apps/main, apps/agent, apps/integrations as needed
npx wrangler secret put ANTHROPIC_API_KEY -c apps/agent/wrangler.jsonc

Generate random secrets with:

openssl rand -hex 32

See the full env table in Reference → Configuration.

Register OAuth apps (only the integrations you want).

See OAuth Apps — separate page because each provider’s dashboard is a maze of its own.

For each, capture the client_id / client_secret / signing_secret and store them as secrets:
Terminal window
```
npx wrangler secret put LINEAR_CLIENT_ID
npx wrangler secret put LINEAR_CLIENT_SECRET
# ... and so on for GITHUB_*, SLACK_*
```
Apply database migrations.
Terminal window
```
npx wrangler d1 migrations apply openma-auth --remote
```
The canonical schema lives at apps/main/migrations/0001_schema.sql.
Deploy.
Terminal window
```
pnpm deploy
```
This runs scripts/deploy.sh, which:
1. Reads service bindings from KV.
2. Generates wrangler.deploy.jsonc with the right services array.
3. Uploads the main worker version.
4. Uploads sandbox worker versions in parallel.
5. Runs a smoke test placeholder.
6. Activates sandbox versions.
7. Activates the main worker version.
Custom domains auto-create DNS records on first deploy (cert provisioning takes ~1 min).

Smoke test.

curl https://app.yourdomain.com/healthz
# → {"ok": true}

# Sign in via the Console
open https://app.yourdomain.com

On first sign-up your tenant is auto-created.

Optional: deploy the docs site

pnpm deploy:docs

Build is ~5 sec, upload is ~10 sec. Routes to docs.yourdomain.com (or whatever you set in apps/docs/wrangler.jsonc).

Optional: staging environment

Every wrangler file has a [env.staging] block. Deploy to staging with:

pnpm deploy --env staging   # main + agent + integrations
pnpm --filter open-managed-agents-docs deploy:staging   # docs

Staging uses separate D1 / R2 / DO ids and routes to *.staging.yourdomain.com.

Where to go next

OAuth Apps Register Linear, GitHub, Slack apps step-by-step.

Operations Custom domains, observability, multi-tenant config, troubleshooting.

Configuration reference Full list of env vars, bindings, schemas.