Deploy Flue Agents With Solo

Flue builds agent workspaces into ordinary HTTP servers. That makes it a useful fit for devopsellence: Flue owns the agent runtime, while devopsellence owns the VM deployment loop, secrets, health checks, logs, rollback, and TLS.

Flue is experimental and its APIs may change. This example was checked against @flue/sdk and @flue/cli 0.3.10; verify the Flue commands locally before using this shape in production.

Build The Flue App

Create a minimal webhook agent:

mkdir my-flue-server
cd my-flue-server
npm init -y
npm install @flue/sdk@0.3.10 @flue/cli@0.3.10 valibot@^1.0.0
mkdir -p .flue/agents

.flue/agents/translate.ts

import type { FlueContext } from '@flue/sdk/client';
import * as v from 'valibot';

export const triggers = { webhook: true };

export default async function ({ init, payload }: FlueContext) {
  const agent = await init({ model: 'openai/gpt-5.5' });
  const session = await agent.session();

  return await session.prompt(`Translate this to ${payload.language}: "${payload.text}"`, {
    result: v.object({
      translation: v.string(),
      confidence: v.picklist(['low', 'medium', 'high']),
    }),
  });
}

Check the Node target before adding devopsellence:

npx flue build --target node
PORT=8080 node dist/server.mjs

From another shell:

curl -fsS http://localhost:8080/health
curl -fsS http://localhost:8080/agents

Containerize It

Dockerfile

FROM node:22-slim AS deps
WORKDIR /app
COPY package.json package-lock.json ./
RUN npm ci

FROM deps AS build
COPY . .
RUN npx flue build --target node

FROM node:22-slim AS runtime
WORKDIR /app
ENV NODE_ENV=production
ENV PORT=8080

COPY package.json package-lock.json ./
RUN npm ci --omit=dev
COPY --from=build /app/dist ./dist

EXPOSE 8080
CMD ["node", "dist/server.mjs"]

cat > .dockerignore <<'EOF'
.git
.env
node_modules
dist
EOF

Add Devopsellence Config

devopsellence init --mode solo

devopsellence.yml

schema_version: 1
organization: solo
project: flue-agent
default_environment: production

build:
  context: .
  dockerfile: Dockerfile
  platforms:
    - linux/amd64

services:
  web:
    ports:
      - name: http
        port: 8080
    healthcheck:
      path: /health
      port: 8080
    env:
      NODE_ENV: production
      PORT: "8080"
    secret_refs:
      - name: OPENAI_API_KEY
        secret: OPENAI_API_KEY

ingress:
  hosts:
    - flue.example.com
  rules:
    - match:
        host: flue.example.com
        path_prefix: /
      target:
        service: web
        port: http
  tls:
    mode: auto
    email: ops@example.com
  redirect_http: true

Store the provider key through devopsellence instead of committing .env:

printf '%s' "$OPENAI_API_KEY" | devopsellence secret set OPENAI_API_KEY --service web --stdin

Deploy

Attach an existing VM:

devopsellence node create prod-1 --host <server-ip-or-hostname> --user root --ssh-key ~/.ssh/id_ed25519
devopsellence agent install prod-1
devopsellence node attach prod-1

Then publish desired state:

devopsellence doctor
devopsellence deploy --dry-run
devopsellence deploy
devopsellence ingress check --wait 5m
devopsellence status

Verify Flue’s generated endpoints and one real agent route:

curl -fsS https://flue.example.com/health
curl -fsS https://flue.example.com/agents
curl -fsS https://flue.example.com/agents/translate/prod-smoke \
  -H "Content-Type: application/json" \
  -d '{"text":"Hello world","language":"French"}'

Node-target Flue sessions are in-memory by default. Add durable session storage inside the Flue app before relying on session history surviving restarts.