Fiscal Clone 3.0

Turbopack-first rebuild of a fiscal.ai-style terminal with Vercel AI SDK integration.

Stack

Next.js 16 App Router
Bun runtime/tooling
Elysia route layer mounted in Next Route Handlers
Turbopack for dev and build
Better Auth (email/password + magic link)
Drizzle ORM (SQLite) + Better Auth Drizzle adapter
Internal API routes via Elysia app module (lib/server/api/app.ts)
Eden Treaty for type-safe frontend API calls
Workflow DevKit Postgres World for background task execution durability
SQLite-backed app domain storage (watchlist, holdings, filings, task projection, insights)
Vercel AI SDK (ai) with dual-model routing:
- Ollama (@ai-sdk/openai) for lightweight filing extraction/parsing
- Zhipu (zhipu-ai-provider) for heavyweight narrative reports (https://api.z.ai/api/coding/paas/v4)

Run locally

bun install
bun run db:generate
bun run db:migrate
bun run dev

Open http://localhost:3000.

The default database path is data/fiscal.sqlite via DATABASE_URL=file:data/fiscal.sqlite.

Production build

bun run db:migrate
bun run build
bun run start

Docker deployment

cp .env.example .env
docker compose up --build -d

For local Docker, host port mapping comes from docker-compose.override.yml (default http://localhost:3000 via APP_PORT). The app calls Zhipu directly via AI SDK for heavy reports and calls Ollama for lightweight filing extraction. When running in Docker and Ollama runs on the host, set OLLAMA_BASE_URL=http://host.docker.internal:11434. Zhipu always targets the Coding API endpoint (https://api.z.ai/api/coding/paas/v4). On container startup, the app applies Drizzle migrations automatically before launching Next.js. The app stores SQLite data in Docker volume fiscal_sqlite_data (mounted to /app/data) and workflow world data in Postgres volume workflow_postgres_data. Container startup runs:

workflow-postgres-setup (idempotent Workflow world bootstrap)
Drizzle migrations for SQLite app tables
Next.js server boot

Docker images use Bun (oven/bun:1.3.5-alpine) for build and runtime.

Coolify deployment

This compose setup is compatible with Coolify as-is (it uses named Docker volumes, not host bind mounts).

Required environment variables in Coolify:

DATABASE_URL=file:/app/data/fiscal.sqlite
BETTER_AUTH_SECRET=<long-random-secret>
BETTER_AUTH_BASE_URL=https://fiscal.b11studio.xyz
BETTER_AUTH_TRUSTED_ORIGINS=https://fiscal.b11studio.xyz
WORKFLOW_TARGET_WORLD=@workflow/world-postgres
WORKFLOW_POSTGRES_URL=postgres://workflow:workflow@workflow-postgres:5432/workflow
Optional: WORKFLOW_POSTGRES_WORKER_CONCURRENCY=10
Optional: WORKFLOW_POSTGRES_JOB_PREFIX=fiscal_

Operational constraints for Coolify:

Keep this service to a single instance/replica. SQLite is file-based and not appropriate for multi-replica shared-write deployments.
Ensure both named volumes are persisted (fiscal_sqlite_data, workflow_postgres_data).
Keep WORKFLOW_POSTGRES_URL explicit so Workflow does not fall back to DATABASE_URL (SQLite).
The app /api/health probes Workflow backend connectivity and returns non-200 when Workflow world is unavailable.

Emergency rollback path:

Set WORKFLOW_TARGET_WORLD=local
Remove/disable WORKFLOW_POSTGRES_URL
Redeploy

Environment

Use root .env or root .env.local:

# leave blank for same-origin API
NEXT_PUBLIC_API_URL=
DATABASE_URL=file:data/fiscal.sqlite
BETTER_AUTH_SECRET=replace-with-a-long-random-secret
BETTER_AUTH_BASE_URL=https://fiscal.b11studio.xyz
BETTER_AUTH_TRUSTED_ORIGINS=https://fiscal.b11studio.xyz

ZHIPU_API_KEY=
ZHIPU_MODEL=glm-4.7-flashx
# optional generation tuning
AI_TEMPERATURE=0.2

OLLAMA_BASE_URL=http://127.0.0.1:11434
OLLAMA_MODEL=qwen3:8b
OLLAMA_API_KEY=ollama
SEC_USER_AGENT=Fiscal Clone <support@fiscal.local>

WORKFLOW_TARGET_WORLD=@workflow/world-postgres
WORKFLOW_POSTGRES_URL=postgres://workflow:workflow@workflow-postgres:5432/workflow
WORKFLOW_POSTGRES_WORKER_CONCURRENCY=10
WORKFLOW_POSTGRES_JOB_PREFIX=fiscal_

# Optional local-world fallback
WORKFLOW_LOCAL_DATA_DIR=.workflow-data
WORKFLOW_LOCAL_QUEUE_CONCURRENCY=100

If ZHIPU_API_KEY is unset, the app uses local fallback analysis so task workflows still run. If Ollama is unavailable, filing extraction falls back to deterministic metadata-based extraction and still proceeds to heavy report generation. ZHIPU_BASE_URL is deprecated and ignored; runtime always uses https://api.z.ai/api/coding/paas/v4.

API surface

All endpoints below are defined in Elysia at lib/server/api/app.ts and exposed via app/api/[[...slugs]]/route.ts.

ALL /api/auth/* (Better Auth handler)
GET /api/health
GET /api/me
GET|POST /api/watchlist
DELETE /api/watchlist/:id
GET|POST /api/portfolio/holdings
PATCH|DELETE /api/portfolio/holdings/:id
GET /api/portfolio/summary
POST /api/portfolio/refresh-prices
POST /api/portfolio/insights/generate
GET /api/portfolio/insights/latest
GET /api/financials/company
GET /api/filings
POST /api/filings/sync
POST /api/filings/:accessionNumber/analyze
GET /api/tasks
GET /api/tasks/:taskId

5.2 KiB Raw Blame History