A Claude Code setup for safer autonomous trading agents
The setup combines , Codex, and a local Gemma model in one autonomous stock-trading system. Each model has a different role, so stronger and more costly models can handle harder decisions while lighter models support smaller tasks.
The system keeps its rules and memory in structured Markdown files such as Charter, Decision Journal, Playbook, and Coaching Log. These files store the mission, decisions, strategy, and so the agent can act consistently over time.
Broker actions use MCP, and real-money trades require human approval before they can run. Safety features include locks, watchdogs, dead-man switches, and heartbeats to reduce duplicate actions, stalled processes, and unsafe automated trades.
Key points
- , Codex, and a local Gemma model are split across different jobs.
- Charter, Decision Journal, Playbook, and Coaching Log files keep goals, choices, strategy, and review notes organized.
- s the system to broker tools, but real-money trades still need human approval.
- Locks, watchdogs, dead-man switches, and heartbeats help prevent unsafe or broken .
- The main value is a safety-first design for complex s.