systemr-cli 2.3.3

System R AI — trading operating system for agents

System R CLI

Trading operating system in your terminal. Like Claude Code, but for trading.

Type natural language. System R runs the full stack — 55 institutional-grade tools, 25 broker adapters, risk engine, G-Score — all from your terminal.

Install

pip install systemr-cli

Or from source:

git clone https://github.com/System-R-AI/systemr-neo.git
cd systemr-neo
pip install -e .

Requires Python 3.11+.

Quick Start

# Set up your profile
systemr setup

# Connect to System R
systemr login

# Start chatting
systemr chat

That's it. Type in plain English:

you > buy TSLA 2% risk

System R will size the position, check risk, show you the trade card, and ask for confirmation before executing.

Commands

Trading

Command	Description
systemr chat	Interactive conversation with System R
systemr size	Position sizing (G-formula)
systemr risk	Pre-trade risk validation
systemr eval	System performance analysis (G-Score)
systemr scan	Market scanner
systemr plan	Trade planning

Local

Command	Description
systemr journal	Trade journal (add/list/show/edit/export)
systemr cron	Scheduled tasks (add/list/remove/run)
systemr doctor	Health check and diagnostics
systemr setup	Profile and rules wizard

Auth

Command	Description
systemr login	Connect to System R
systemr logout	Disconnect
systemr whoami	Show current session

Chat Commands

Inside systemr chat, use slash commands:

Command	Action
/morning	Morning briefing (4 parallel agents)
/eod	End of day review + journal
/plan	Plan today's trades
/portfolio	Show open positions
/risk	Risk dashboard
/memory <query>	Search past memories
/sessions	List recent sessions
/cron	Manage scheduled tasks
/permissions	View/switch safety profile
/remember <text>	Save a memory
/model <name>	Switch LLM model
/credits	Session credit usage
/help	All commands

Safety

Three-tier confirmation protocol:

• AUTO — Read-only actions (quotes, analysis, calculations)
• CONFIRM — Trade actions (place/cancel/modify order) — show details, y/n
• DOUBLE_CONFIRM — Destructive actions (kill switch) — type "KILL"

Permission profiles: paper (max safety), standard (default), experienced (relaxed stops).

Architecture

~/.systemr/
  PROFILE.md          # Trader identity, risk params
  RULES.md            # Hard/soft rules + standing orders
  auth.json           # API credentials (chmod 600)
  config.json         # Settings, model config
  journal.db          # SQLite trade journal
  memory/
    MEMORY.md         # Index
    YYYY-MM-DD.md     # Daily trading logs
    *.md              # Lessons, notes, violations
  sessions/
    last_session.json # Resume state
  cron/
    jobs.json         # Scheduled tasks
    runs/             # Run history (JSONL)

All data stays on your machine. The backend (agents.systemr.ai) processes requests — your profile, rules, and journal never leave your disk.

Backend

System R CLI connects to agents.systemr.ai:

• 55 MCP tools (sizing, risk, intelligence, planning, behavioral)
• 25 broker/exchange adapters
• 187 domain services
• Per-agent encryption (AES-128-CBC)
• Multi-model LLM via AWS Bedrock (Claude, GPT, Nova)
• Compute credit billing (USDC, SOL, OSR)

Development

git clone https://github.com/System-R-AI/systemr-neo.git
cd systemr-neo
pip install -e ".[dev]"
pytest

Rules

• Decimal for all money — never float
• structlog only — zero print()
• Additive-only — never remove/rename
• Tests for every module

Stats

• 33 source files, ~8,800 lines
• 206 tests passing
• Zero bare print() statements
• Python 3.11+

Links

• systemr.ai — Documentation
• agents.systemr.ai — API
• app.systemr.ai — Portal

Built by Ashim Nandi and Shannon at System R AI.