dr-trading-system

Treat this as a framework skill, not a ready-to-trade bot.

Use it to help the user set up or refine a local trading-system deployment with:

• - provider configuration
• watchlists
• strategy configs
• job configs
• paper-mode execution flows
• approval-gated buy/sell handling
• daily assessment reports

Default workflow

1. 1. Confirm whether the user wants a new setup, an edit, or an audit.
2. If setup is needed, run the conversational wizard one question at a time.
3. Summarize the planned deployment before generating files.
4. Generate local deployment files from the examples and references.
5. Keep secrets, credentials, runtime state, generated reports, schedules, and deployment-specific configs outside the reusable skill package.
6. Default to paper-first and approval-gated operation unless the user explicitly asks otherwise.

Conversational wizard

If the user asks to set up the system, guide them through the setup conversationally.

Use this exact activation phrase when helpful:

INLINECODE0

The wizard is conversation-driven. There is no separate UI or /wizard command.

Ask questions one at a time. At minimum, cover:

• - market and broker/provider
• deployment goal
• strategy count and first strategy type
• watchlists
• daily assessment vs trusted paper mode
• approval rules for buys and sells
• local file/output locations

Before generating files, provide a short deployment summary and get confirmation if the setup has material risk or ambiguity.

Recommended first rollout

For a first deployment, prefer:

• - 1 strategy
• 1 market
• 2 watchlists
• paper mode only
• buy and sell approval required
• daily assessment mode first

Reference map

Read only what is needed:

• - references/install-and-setup.md — install/use flow and local deployment boundaries
• INLINECODE3 — high-level scope and operating model
• INLINECODE4 — reusable engine structure and responsibilities
• INLINECODE5 — provider adapter interface and expectations
• INLINECODE6 — how to run the setup conversation
• INLINECODE7 — wizard structure and setup scope
• INLINECODE8 — question order and branching
• INLINECODE9 — expected local files/artifacts
• INLINECODE10 — safe operating patterns
• INLINECODE11 — example provider, strategy, watchlist, job, and report configs

Boundaries

Keep these local to the deployment, not in the reusable skill:

• - secrets
• provider credentials
• provider session info
• runtime state
• generated reports
• local schedules
• machine-specific paths
• deployment-specific watchlists/jobs

Do not present paper or live outputs as trustworthy until provider freshness, permissions, completeness, and report sanity have been validated.