Environment Variables

Configure Uranus with environment variables.

Configuration Methods

wrangler.toml

[vars]
ENVIRONMENT = "production"
BROWSERBASE_PROJECT_ID = "your-project-id"

Cloudflare Dashboard

1. Go to Workers & Pages
2. Select your worker
3. Settings > Variables
4. Add environment variables

Secrets (Sensitive Data)

# Add secret via CLI
npx wrangler secret put BROWSERBASE_API_KEY

Core Variables

ENVIRONMENT

Runtime environment identifier.

Value	Description
development	Local development
staging	Staging environment
production	Production deployment

[vars]
ENVIRONMENT = "production"

Browser Automation

BROWSERBASE_PROJECT_ID

Browserbase project identifier for browser automation.

[vars]
BROWSERBASE_PROJECT_ID = "proj_xxxxx"

BROWSERBASE_API_KEY

Browserbase API key (use secrets).

npx wrangler secret put BROWSERBASE_API_KEY

Authentication

CF_ACCESS_AUD

Cloudflare Access audience tag for JWT validation.

[vars]
CF_ACCESS_AUD = "your-access-audience-tag"

AI Providers

Store API keys as secrets:

OpenAI

npx wrangler secret put OPENAI_API_KEY

Anthropic

npx wrangler secret put ANTHROPIC_API_KEY

Resource Bindings

D1 Database

[[d1_databases]]
binding = "DB"
database_name = "agents-dashboard"
database_id = "your-database-id"

R2 Storage

[[r2_buckets]]
binding = "BUCKET"
bucket_name = "agents-dashboard-files"

Vectorize

[[vectorize]]
binding = "VECTORIZE"
index_name = "agents-dashboard-embeddings"

Durable Objects

[[durable_objects.bindings]]
name = "AGENT"
class_name = "DashboardAgent"

Queues

[[queues.producers]]
binding = "WORKFLOWS_QUEUE"
queue = "agents-dashboard-workflows"

Browser Rendering

[browser]
binding = "BROWSER"

The default browser provider (BROWSER_PROVIDER=cf) drives this binding with @cloudflare/puppeteer; Steel is only needed for hitlProvider = "steel" sessions.

AI Gateway

[vars]
AI_GATEWAY_ID = "your-gateway"

Partner models — the Jev typed-decision model (typesafe/jev) — bill through AI Gateway and are always invoked through the configured gateway. Without it, every Jev-backed decision takes its non-partner fallback (see Typed Decisions).

Agent Memory (optional, private beta)

When Cloudflare's Agent Memory binding is present as AGENT_MEMORY, settings.memory.provider = "cloudflare" uses it; otherwise the local provider is used regardless of the setting.

Platform API secret

PLATFORM_INTERNAL_SECRET (alias: OAC_FRONTDOOR_SECRET) is the bearer products use on /api/internal/v1/*; OAC_API_BASE is where the manage worker reaches OAC's own API. See Supporting Services → Platform API.

Local Development

.dev.vars

Create .dev.vars for local secrets:

BROWSERBASE_API_KEY=your-dev-key
OPENAI_API_KEY=sk-dev-xxxxx

Note: Add .dev.vars to .gitignore

wrangler.toml [env.development]

[env.development]
vars = { ENVIRONMENT = "development" }

[env.development.d1_databases]
[[env.development.d1_databases]]
binding = "DB"
database_name = "agents-dashboard-dev"
database_id = "your-dev-database-id"

Environment-Specific Config

Development

[env.development]
name = "uranus-dev"

[env.development.vars]
ENVIRONMENT = "development"

Staging

[env.staging]
name = "uranus-staging"

[env.staging.vars]
ENVIRONMENT = "staging"

Production

[env.production]
name = "uranus"

[env.production.vars]
ENVIRONMENT = "production"

Deploy to Environment

# Development
npx wrangler deploy --env development

# Staging
npx wrangler deploy --env staging

# Production (default)
npx wrangler deploy

Variable Reference

Variable	Required	Description
ENVIRONMENT	Yes	Runtime environment
BROWSERBASE_PROJECT_ID	No	Browser automation project
BROWSERBASE_API_KEY	No	Browser automation key
CF_ACCESS_AUD	No	Cloudflare Access audience

Secrets Management

Add Secret

npx wrangler secret put SECRET_NAME

List Secrets

npx wrangler secret list

Delete Secret

npx wrangler secret delete SECRET_NAME

Secrets in CI/CD

- name: Set secrets
  run: |
    echo "${{ secrets.BROWSERBASE_API_KEY }}" | npx wrangler secret put BROWSERBASE_API_KEY

Best Practices

1. Use Secrets for Sensitive Data

Never put API keys in wrangler.toml:

# Good
npx wrangler secret put API_KEY

# Bad - never do this
[vars]
API_KEY = "sk-xxxxx"

2. Environment Separation

Use different resources per environment:

[env.development.d1_databases]
database_name = "agents-dashboard-dev"

[env.production.d1_databases]
database_name = "agents-dashboard"

3. Document Variables

Keep a reference of required variables:

# Required Variables
ENVIRONMENT - Runtime environment
DB - D1 database binding
BUCKET - R2 bucket binding

# Optional Variables
BROWSERBASE_* - Browser automation
CF_ACCESS_AUD - SSO authentication

4. Validate on Startup

Check required variables:

if (!env.DB) {
  throw new Error('DB binding is required')
}