Installation

Clone the workspace

git clone https://github.com/dlepaux/gordon-workspace.git
cd gordon-workspace

All service repos are siblings inside the workspace directory, not submodules.

First-run bootstrap

make dev-init

dev-init is idempotent. It:

• Checks required tools (Docker, Rust, Node, Python, overmind, sccache, cargo-nextest)
• Initialises a local Postgres cluster at ./data/postgres/
• Runs gordon-migrate to apply all schema migrations
• Creates the NATS JetStream stream (gordon-bus)
• Writes .env.dev with dev-safe defaults

Run this once per machine. Re-running is safe.

Start the dev stack

make dev-up

This launches the full stack via overmind (Procfile.dev):

Process	What it runs
postgres	Host Postgres (not Docker)
nats	Host NATS server with JetStream
migrate	gordon-migrate one-shot — exits after applying migrations
data	gordon-data on :8081
risk	gordon-risk on :8082
manager	gordon-manager on :8083
bot	gordon-bot on :8084
executor	gordon-executor on :8085
console	gordon-console (Next.js) on :3000

Services auto-restart until migrate completes and the least-privilege roles exist.

Tail a specific process:

make dev-logs P=data     # gordon-data logs
make dev-logs P=manager  # gordon-manager logs

Check process status:

make dev-status

Seed market data

make dev-seed

Backfills BTC+ETH 1-year of 1m klines, higher-TF precompute, GEX, overlays, and macro data. First run takes approximately 20 minutes. The command is idempotent — re-running only fetches data newer than what is already in the database.

Verify

Open http://localhost:3000 in your browser. gordon-console should show the operator dashboard. If any service is unhealthy, check its logs:

make dev-logs P=executor

Stop the stack

make dev-down   # stop processes, preserve ./data/
make dev-reset  # stop AND wipe ./data/ (full reset — re-run dev-init after)

Connect to the dev database

make dev-psql   # opens psql shell

\dt             -- list tables
SELECT count(*) FROM market_data.spot_klines;
\q

Next step

With the stack running and data seeded, you are ready to run your first backtest.