TradingCodex: An Evolving Multi-Agent Investment Team for Codex

Dynamics · Statics · Install · User Guide

TradingCodex brings an evolving investment team into native Codex. Ask a question in ordinary language; the right agents collaborate, challenge the evidence, and leave behind reusable context for the next decision.

It is research-first, local-first, and paper-first. A research answer never becomes a broker action on its own.

Dynamics

The team changes with the question

TradingCodex does not send every ticker through the same checklist. Head Manager reads the mandate, chooses the smallest useful group of agents, runs independent work in parallel when useful, and revises the team when accepted evidence exposes a new gap or conflict.

The agents

The workflow is a loop

question
  -> Head Manager
  -> smallest useful specialist wave
  -> accepted evidence and artifacts
  -> follow up / add a role / request review / synthesize / stop
  -> next question or decision

The next agent is chosen by the next useful question, not by a preset research DAG. A narrow question can stay direct. A high-consequence recommendation can expand to independent portfolio, risk, and judgment review. There is no execution agent.

Statics

The context that stays with you

The current workflow is dynamic. The useful work it produces is static, readable, and reusable: files and memory that make the next analysis better without hiding the evidence inside a chat or silently changing system rules.

The workspace stays inspectable and versionable:

trading/
  research/       # snapshots, datasets, artifacts, calculations
  reports/        # research reports and handoffs
  forecasts/      # forward-looking judgments
  decisions/      # decision context and review history
wikis/            # local and active Wiki knowledge
investment-brains/  # user-owned inquiry frameworks

Knowledge, Brain, Strategy, and Decision Memory remain distinct. A Wiki does not become proof, a Brain does not choose agents, and an improvement record does not silently rewrite prompts, skills, policy, or execution gates.

Install

For normal use, attach TradingCodex to the folder where you want research to live. Do not clone this source repository into that workspace.

You need Git, uvx, and an installed, authenticated codex CLI.

On macOS/Linux:

cd /path/to/an/empty-workspace
uvx --refresh --from tradingcodex tcx attach . && ./tcx doctor
./tcx service ensure

On native Windows PowerShell:

cd C:\path\to\an\empty-workspace
uvx --refresh --from tradingcodex tcx attach .
.\tcx.cmd doctor
.\tcx.cmd service ensure

After installation, fully restart Codex, reopen and trust the workspace, and trust every TradingCodex hook when Codex prompts you. If hooks are presented one at a time, approve each project hook before starting a new task. The viewer URL is printed by ./tcx service status; the release default is http://127.0.0.1:48267/. End users do not need Node, npm, or a separate frontend server.

The Head Manager inherits the active Codex model and reasoning setting. For full research workflows, we recommend gpt-5.6-sol with high or xhigh reasoning.

Start with skills

Start with the skill that matches the work. Each example is a prompt you can paste into a new Codex task; the full skill directory has exact invocation rules, boundaries, and deeper examples.

Use the viewer to inspect the resulting Episodes, Library, Wiki, and System posture. It is read-only; use Codex to continue, narrow, or stop the work.

The guardrail

Research, Wiki, Brain, Strategy, memory, and automation do not imply broker authority. Provider connection, ticket drafting, checks, approval, and final action remain separate checkpoints. Paper is built in; live providers require their own installation, policy approval, explicit confirmation, sync, and audit gates.

Raw credentials never belong in prompts, workspace files, reports, API/MCP responses, artifacts, or audit data.

Learn more

• Dynamic workflow — evidence-driven agent selection, reassessment, and stopping.
• Knowledge Wiki — reusable background knowledge and explicit write boundaries.
• Investment Brain — inquiry and interpretation frameworks.
• Improve — Decision Memory, postmortems, lessons, and controlled evolution.
• Installation — updates, runtime homes, service startup, and recovery.
• Safety, policy, and execution — permissions, approvals, brokers, secrets, and final-action boundaries.
• Documentation index — the complete product reference.

Developing TradingCodex

This repository is product source, not a user workspace. Start with CONTRIBUTING.md, then follow the validation route in docs/validation-and-test-plan.md.

License

TradingCodex is an Apache-2.0 open-core project. See LICENSE, NOTICE, and TRADEMARKS.md.