execute-trade

name	execute-trade
description	Orchestration skill for Ratatosk trade execution. Thesis → Council → Risk Check → Execute → Audit → Report. End-to-end trade lifecycle.

Execute Trade (Orchestration — Ratatosk)

Overview

Chains the ADLC Build Loop adapted for investment operations. From market signal to executed trade with full audit trail.

When to Use

• Ratatosk has identified a trade opportunity
• Trade thesis is ready for evaluation
• End-to-end trade execution pipeline

The Sequence

Step 1: Trade Thesis (structured from market signal)
Step 2: STRIDE on trade execution
Step 3: Eval Council (HEAVY — investment-adapted)
Step 4: Risk Gate (HARD — position/drawdown/budget limits)
Step 5: Execute Trade (exchange API)
Step 6: Audit Log
Step 7: Stop Slop on report
Step 8: Report (Telegram)
Step 9: Feedback capture (post-settlement)

Step 1: Trade Thesis

• Skill: trade-thesis (Ratatosk domain PRD)
• Structure signal into: market context, conviction rationale (quantified), risk parameters, exit criteria, position sizing
• Reject ambiguous signals (no vague directional calls)

Step 2: Security Review

• Skill: security-review (STRIDE mode — trade execution)
• Spoofing: order spoofing, fake signals
• Tampering: parameter modification
• Repudiation: decision audit trail
• Info Disclosure: API key/position exposure
• DoS: exchange rate limiting
• Elevation: unauthorized mode escalation

Step 3: Eval Council — Full

• Skill: eval-council (HEAVY — 6 investment-adapted personas)
• Architect (Portfolio Strategy): Portfolio fit, correlation risk, concentration
• Skeptic (Bearish Thesis): Bear case, thesis inversion, what if wrong?
• Operator (Execution Feasibility): Liquidity, slippage, timing, market hours
• Executioner (Agent Capability): All parameters specified? Can agent execute?
• Security Auditor (Funds Safety): API exposure, position sanity, mode authorization
• First Principles (Thesis Soundness): Data-driven? Edge identified? Sound reasoning?

Step 4: Risk Gate (HARD)

This is a non-negotiable automated gate. No human override for limit breaches.

Check	Limit	Action on Breach
Position size	Per-strategy max	BLOCK trade
Portfolio concentration	Max % in single asset	BLOCK trade
Daily drawdown	Max daily loss limit	BLOCK all trading
Budget	Remaining budget check	BLOCK trade
Calibration	Strategy calibration score	Reduce size or BLOCK

Step 5: Execute Trade

• Place order via exchange API
• Capture: order ID, entry price, size, timing, exchange, order type
• Set exit criteria (stop loss, take profit, time-based exit)

Step 6: Audit Log

Full structured audit entry:

{
  "pipeline_run_id": "uuid",
  "trade_id": "...",
  "thesis_summary": "...",
  "council_verdict": "approved",
  "risk_gate": "passed",
  "entry": { "price": 0, "size": 0, "time": "ISO8601", "exchange": "..." },
  "exit_criteria": { "stop_loss": 0, "take_profit": 0, "time_exit": "ISO8601" },
  "stride_summary": "...",
  "calibration_score": 0
}

Step 7: Stop Slop

• Skill: stop-slop (content mode on trade summary)
• Score the trade report on 5 dimensions
• Threshold: 35/50

Step 8: Report

• Deliver via Telegram
• Include: trade summary, thesis, risk parameters, council verdict

Step 9: Feedback Capture (Post-Settlement)

• After trade settles: predicted P&L vs actual
• Classify: thesis error, timing error, sizing error, execution error
• Feed into calibration and experiment loop
• Update signal weighting based on outcome

Escalation

Trigger	Action
Risk limit breach	Immediate notification to Eric
Council BLOCKED verdict	Trade cancelled, thesis logged for review
Execution failure	Fix loop triggered, position check initiated
Loss exceeds 2x expected	Post-mortem triggered, calibration review