Skip to content

Latest commit

 

History

History
88 lines (57 loc) · 2.64 KB

File metadata and controls

88 lines (57 loc) · 2.64 KB

Crucible CIO Team — Codespace Quickstart

Your Codespace has automatically installed all Python dependencies, Claude Code, and initialized the SQLite database. Follow these steps to get operational.


Step 1 — Configure Environment Variables

Copy the environment template and add your API keys:

cp .env.template .env

Open .env and fill in:

ANTHROPIC_API_KEY=sk-ant-...      # Required — get from console.anthropic.com
FRED_API_KEY=...                   # Required — get from fred.stlouisfed.org/docs/api/api_key.html
IB_HOST=127.0.0.1                  # Optional — only if connecting to Interactive Brokers
IB_PORT=7497
IB_CLIENT_ID=1

Step 2 — Verify FRED Connection

Confirm your FRED API key is working:

python scripts/verify-fred.py

Expected output: FRED connection OK — [indicator name] retrieved successfully.

If you see an error, double-check your FRED_API_KEY in .env.


Step 3 — Open Claude Code

claude

Claude Code will open in the integrated terminal. All 40+ slash commands are immediately available.


Step 4 — Configure Your Fund

Run the interactive onboarding wizard. This populates context/fund-mandate.md, context/portfolio-state.md, and context/risk-limits.md — required inputs for every agent.

/setup

The wizard conducts a five-stage interview: Fund Identity, Strategy, Risk Parameters, Operations, and Regulatory. Takes about ten minutes. Nothing works correctly until this is complete.


Step 5 — Run Your First Pipeline

Once setup is complete, submit your first paper trade through the full agent stack:

/run-pipeline Long 2% NAV ES futures, trend signal fires on 20-day breakout

The pipeline will run all 11 agents in sequence, apply the context bus, and produce a unified Pipeline Report with a final actionable instruction.


Available Commands (once setup is complete)

Command What it does
/run-pipeline [decision] Full pipeline — primary entry point for all trade decisions
/stress-test Run portfolio through 5 historical crisis scenarios
/calibration-report Monthly agent accuracy and calibration metrics
/export-audit Compile 30-day audit document for LP review
/crucible [trade] Five-agent panel review (compliance, risk, macro, signal, systems)
/postmortem [trade] | [outcome] | [date] Feed losing trade back through agents to find what was missed
/regime-classifier Current market regime state machine
/drawdown-monitor Portfolio drawdown circuit breaker status

See CLAUDE.md for the full command reference.