Standalone Operator Guide (No AI)
This guide is for operators who want the b1e55ed data + signal engine only.- Pure CLI (plus optional dashboard)
- No AI dependency
- Works on a single VPS
What you get
- Signal engine: producers → events → brain synthesis
- REST API (
/api/v1) for health, producers, oracle, etc. - Dashboard (optional, but useful): local web UI
- CLI control plane: setup, brain cycles, health checks, positions, contributors
- Systemd-ready: run API + brain loop as services
Prerequisites
Minimum server
- VPS: 2 vCPU / 2 GB RAM minimum (4 GB recommended)
- OS: Ubuntu 22.04+ (any Linux is fine; commands below assume Ubuntu)
- Disk: 20 GB
Accounts / access
- A GitHub account (needed because the installer pulls from GitHub)
Software
- Python 3.11+ (the installer will verify;
uvmanages the venv/tool) curl
Install
Run the installer as the user that will operate the engine (recommended: a dedicatedb1e55ed user):
b1e55ed: command not found
Fix:
First run (wizard)
Run:- Identity forge (creates your contributor identity)
- Universe + config (writes
config/user.yaml) - Producer registration (enables/disables producers and schedules)
- First brain run (verifies end-to-end ingestion → synthesis → storage)
- API setup (so you can query health/oracle and use the dashboard)
What the 0xb1e55ed prefix means
You will see 0xb1e55ed in banners and identity output.
- It’s the system’s blessed hex prefix and a human-visible marker that you’re interacting with the correct toolchain.
- It also shows up in identity / provenance contexts so operators can spot “wrong box / wrong env” mistakes.
How long does identity forge take?
The forge step is CPU-bound. Rough ballpark:| Hardware | Typical forge time |
|---|---|
| 2 vCPU shared VPS | 30–120 seconds |
| 4 vCPU VPS | 15–60 seconds |
| Modern laptop CPU | 5–30 seconds |
b1e55ed identity --help.)
⚠️ Common failure point #2: forge feels “stuck”
What to do:
Start the engine
Start API + dashboard (recommended)
- API:
http://127.0.0.1:5050(health:/api/v1/health) - Dashboard:
http://127.0.0.1:5051
Run the brain (signal loop only)
For a single cycle:Core CLI commands
These are the commands you’ll use daily.| Command | What it does |
|---|---|
b1e55ed wizard | First-run interactive setup (identity, config, producers, first run, API) |
b1e55ed start | Starts API + dashboard together (ports 5050/5051 by default) |
b1e55ed brain | Runs one brain cycle (add --full for slower producers) |
b1e55ed scan | Not a top-level command in current releases. Use b1e55ed alerts (recent alerts) + dashboard views to “scan” system state. |
b1e55ed health | Cron-safe health check (useful for monitoring) |
b1e55ed status | Human-readable system status summary |
b1e55ed signal | Inject operator intel (curator signal) |
b1e55ed positions | Lists open positions with best-effort PnL |
b1e55ed contributors | View/manage contributor reputation + provenance |
--help. For the full list, see CLI reference.
Config tuning
Your operator config is at:config/user.yaml
- the universe (symbols)
- execution mode (paper vs live)
- producer enablement + weights (how much each producer influences synthesis)
Key fields to know
(Exact keys are fully documented in Configuration; below is the operator-focused subset.)| Area | What to look for | Why it matters |
|---|---|---|
| Universe | universe.symbols | What assets get collected/scored |
| Execution | execution.mode (paper/live) | Controls whether execution is real |
| Producers | producers.enabled / producers.weights | Decide what data sources contribute to conviction |
Sample config: BTC-focused
Sample config: multi-asset (top 10)
b1e55ed start).
⚠️ Common failure point #3: YAML indentation / syntax errors
Fix:
Producer reference (13 core producers)
b1e55ed uses “producers” to collect different kinds of signals. This table is designed for operators deciding what to enable first.| Producer | Domain | What it does | When to enable | Cost / complexity |
|---|---|---|---|---|
technical-analysis | technical | Generates TA features/signals (trend, momentum, etc.) | Baseline; usually on | Low |
price-alerts | technical | Watches price moves and emits alert-like signals | If you want fast reaction / alerts | Low |
orderbook-depth | technical | Pulls liquidity/imbalance metrics from an external orderbook endpoint | When you have reliable orderbook data source | Medium (needs endpoint) |
market-events | events | Emits structured market events used by the system | Default-on for context | Low |
onchain-flows | onchain | On-chain flow/transfer features | When operating liquid large caps / majors | Medium (API/data deps) |
whale-tracking | onchain | Tracks large entity movements | When you care about large flow narratives | Medium |
stablecoin-supply | onchain | Monitors stablecoin supply / issuance patterns | Macro/liquidity regime sensitivity | Medium |
tradfi-basis | tradfi | TradFi basis / funding proxies | If you trade BTC/ETH and care about leverage crowding | Medium |
etf-flows | tradfi | ETF flow signals (where applicable) | If you care about US market structure flows | Medium |
market-sentiment | social | Sentiment signal from supported sources | If you want risk-on/off overlay | Medium |
social-intel | social | Social narrative / intel ingestion | When you want discretionary “what’s being talked about” | Medium |
curator-intel | curator | Operator-injected signals (manual intel) | Always if humans provide inputs | Low |
ai-consensus | curator | AI consensus synthesis (requires AI access) | Standalone operators usually leave this off | High (external AI) |
- This guide excludes the internal
templateproducer (used as a scaffold). - Producers have schedules; the wizard helps register/enable them.
Running as a service (systemd)
Below are copy/paste unit files that run:b1e55ed-api.service: API server (port 5050)b1e55ed-brain.service: a simple loop that runsb1e55ed brainevery 5 minutes
- You installed as the
b1e55eduser (recommended) b1e55edis available at/home/b1e55ed/.local/bin/b1e55ed
1) API service
2) Brain loop service
3) Enable + start
Oracle setup (optional)
The oracle is a read-only endpoint that exposes producer provenance for agents and other consumers. If you are running standalone, you typically only need this if:- you want external systems to query provenance, or
- you plan to add an agent layer later.
- Ensure your API is reachable (either via Tailscale or a reverse proxy)
- Use the oracle endpoint: