prediction-market-bot-dawn（预测市场机器人黎明）

Run a Dawn strategy lifecycle

Goal

Execute a complete dawn strategy workflow: install/check CLI, authenticate, prepare funding, create and iterate strategy code, launch paper/live runs, monitor status, and stop safely when requested.

When to use

Use this skill when the user asks to:

• create a strategy from plain-English intent,
• revise or upload strategy code,
• launch paper or live strategy runs,
• monitor run health/positions/logs,
• stop or debug an active run.

Install and preflight

Install dawn if needed:

npm install -g @dawnai/cli

Verify:

# Print current Dawn CLI version
dawn version
dawn --help

Local source workflow only:

cd dawn-cli
npm install
npm run build
./install.sh

Command map

Auth:

• dawn auth login
• dawn auth status
• dawn auth logout

Account:

• dawn account overview
• dawn account fund
• dawn account wallet

Strategy authoring:

• dawn strategy list
• dawn strategy create "<text>"
• dawn strategy status <conversationId>
• dawn strategy revise <conversationId> "<text>"
• dawn strategy rules <conversationId> list
• dawn strategy rules <conversationId> approve <rule-index>
• dawn strategy rules <conversationId> approve-all
• dawn strategy code <conversationId> status
• dawn strategy code <conversationId> generate
• dawn strategy code <conversationId> export [--out <path>] [--json]
• dawn strategy code <conversationId> upload <path-to-file>

Launch and operations:

• dawn strategy launch <conversationId> --budget <usd> [--live] [--hours N]
• dawn strategy positions <conversationId> [--strategy-id <strategyId>]
• dawn run list
• dawn run status <conversationId>
• dawn run logs <conversationId> [--limit N]
• dawn run stop <conversationId>

Standard flow

1. Authenticate: dawn auth login.
2. Confirm funding path: dawn account fund (required for live runs).
3. Create strategy: dawn strategy create "<request>" and capture conversationId.
4. Iterate strategy:

- revise prompt (strategy revise) and/or upload files (strategy code ... upload), - review/approve rules, - generate code, - export code when needed (--json for multi-file map).

1. Launch:

- paper: dawn strategy launch <conversationId> --budget 50 - live: dawn strategy launch <conversationId> --budget 50 --live - custom duration: add --hours N

1. Monitor:

- dawn run status <conversationId> - dawn strategy positions <conversationId> - dawn run logs <conversationId> --limit N

1. Stop when requested: dawn run stop <conversationId>, then verify status again.

Monitoring loop

For active monitoring sessions:

1. Query dawn run status <conversationId>.
2. Record timestamp, isRunning, status, and active strategy IDs.
3. Query dawn strategy positions <conversationId> for holdings/PnL.
4. Query dawn run logs <conversationId> --limit N for execution details.
5. If records look stale or missing, wait briefly and retry once.

Troubleshooting

• "Not authenticated. Run: dawn auth login": run dawn auth login and retry.
• Auth callback completes but CLI appears stuck: interrupt once and retry login.
• "No strategy version found...": create/revise/upload strategy code, then relaunch.
• "No strategies found for this agent" on stop: verify conversationId, then check dawn run status.
• Live launch fails: re-check funding path with dawn account fund.

Run checklist

Dawn Strategy Runbook
- [ ] Preflight complete
- [ ] Auth complete
- [ ] Funding path checked (or user confirmed paper-only)
- [ ] conversationId captured
- [ ] Strategy code generated/uploaded
- [ ] Launch run completed (paper/live)
- [ ] strategyId captured (if launched)
- [ ] Monitoring snapshots collected
- [ ] Stop executed (if requested)
- [ ] Final status verified

Skills

Individual skills for each command:

Required output

When using this skill, always return:

• conversationId,
• strategyId (if launched),
• run mode (paper/live),
• latest monitoring summary,
• exact next command to run (or the last command run).