Deploy your own AEGIS agent on Cloudflare Workers in under 10 minutes.
- Node.js 18+
- A Cloudflare account (free tier works)
- An Anthropic API key (for Claude)
- Optional: Groq API key (free tier available, used for fast classification)
git clone https://github.com/Stackbilt-dev/aegis-oss.git
cd aegis-oss/web
npm installCopy the example config and fill in your Cloudflare account details:
cp wrangler.toml.example wrangler.tomlEdit wrangler.toml:
- Uncomment and set
account_idto your Cloudflare account ID - Set a
database_namefor your agent (or keepmy-agent)
npx wrangler d1 create my-agentThis prints a database_id — paste it into wrangler.toml under [[d1_databases]].
Then run the schema migration:
npx wrangler d1 execute my-agent --file=schema.sqlcp src/operator/config.example.ts src/operator/config.ts
cp src/operator/persona.example.ts src/operator/persona.tsEdit config.ts to set your name, persona traits, and which integrations to enable. See Configuration for the full reference.
At minimum, you need an auth token and an AI model key:
npx wrangler secret put AEGIS_TOKEN # A random bearer token you'll use to authenticate
npx wrangler secret put ANTHROPIC_API_KEY # Your Claude API keyOptional secrets for additional capabilities:
npx wrangler secret put GROQ_API_KEY # Fast classification (Llama 3.3 70B)
npx wrangler secret put GITHUB_TOKEN # Repository scanning, issue management
npx wrangler secret put BRAVE_API_KEY # Web research
npx wrangler secret put RESEND_API_KEY # Email notificationsnpx wrangler deployYour agent is now live at https://your-worker-name.your-subdomain.workers.dev.
Visit your worker URL. You'll be prompted for a bearer token — use the AEGIS_TOKEN you set in step 5.
Check the health endpoint:
curl https://your-worker.workers.dev/healthYou should see the version number and system status.
For local development without deploying:
# Create a .dev.vars file with your secrets
cat > ../.dev.vars << 'EOF'
AEGIS_TOKEN=test-token
ANTHROPIC_API_KEY=sk-ant-...
GROQ_API_KEY=gsk_...
EOF
# Run locally
npx wrangler dev- Architecture — understand the cognitive kernel
- Configuration — customize your agent's behavior
- Memory System — how AEGIS remembers and learns