Schedule a weekly watchlist research review

TL;DR

For a saved watchlist, choose its weekly AI research brief in notification preferences. A general LaunchAgent schedule is a separate feature with an explicit scope and prompt. The recorded AAPL/MSFT example saved a disabled Friday schedule and started no Aurora run. Its 09:30 Eastern setting is a fifty-nine-minute eligibility window, not a guaranteed start time. Review cadence, cooldown and future credit use before enabling, then inspect the resulting run history.

Follow one worked path

Step 1 of 3

Inspect the saved inactive schedule

Read the saved AAPL/MSFT schedule below. Check its $0 starting balance, Friday window and disabled state.

Research scope

AAPL and MSFT, public filings only

State

Inactive $0 paper portfolio; 1 LaunchAgent rule

Schedule

Friday, 09:30 Eastern; 59-minute eligibility window

Evaluation frequency

Constant

Cooldown

1,440 minutes

Iterations

At most 10 per future run

Activation / execution

No active schedule or current agent run

What you should see

The recorded readback is type paper, isActive false, with one disabled schedule, zero active and zero running. It started no Aurora run.

Step 2 of 3

Create an inactive research schedule

For the weekly example, keep active false and inspect the exact prompt, Friday time window, cooldown and maximum iterations before saving.

What you should see

A disabled schedule is not a running agent. The recorded example remained inactive and started no Aurora run.

Review the inactive schedule in Aurora →

Step 3 of 3

Inspect the next artifact

Open Agents and inspect the saved scope, Friday eligibility window, cooldown and inactive state. Enable your own reviewed schedule only when you choose to proceed; read its future run history there.

Research scope

AAPL and MSFT, public filings only

State

Inactive $0 paper portfolio; 1 LaunchAgent rule

Schedule

Friday, 09:30 Eastern; 59-minute eligibility window

Evaluation frequency

Constant

Cooldown

1,440 minutes

Iterations

At most 10 per future run

What you should see

The captured example is a saved inactive schedule. Its future research output remains uncaptured.

An actual saved schedule with research off

We saved a dedicated $0 scheduled research portfolio in NexusTrade on October 10, 2026 with active false. get_portfolio returned type paper and isActive false. A filtered list_scheduled_agents read showed one matching disabled schedule, zero active and zero running. No Aurora run was started.

This example reviews a fixed AAPL/MSFT list. Native watchlist AI research briefs and LaunchAgent schedules are separate features. For your saved watchlist, use its notifications preferences; for a general research agent, inspect a complete schedule and prompt like the one below.

Create an inactive research schedule

Call create_scheduled_trading_agent with the JSON above. Leave active false while checking it. Omit portfolio_id to create a dedicated example rather than modify a working portfolio. Creation costs no LLM credits; enabling it can launch future Aurora work that consumes credits.

The weekly condition is eligible on Friday from 09:30 until just before 10:29 Eastern time. It checks under Constant evaluation. This is a time window, not a guarantee of execution at exactly 09:30; outages or an inactive portfolio can leave it without a run. The 1,440-minute cooldown limits repeat launches.

{
  "portfolio_name": "Public SEO example - inactive Friday research",
  "strategy_name": "Friday source review",
  "initial_value": 0,
  "active": false,
  "deployment_frequency": "Constant",
  "schedule": {
    "frequency": "weekly",
    "day_of_week": 5,
    "time": "09:30"
  },
  "initial_message": "Research AAPL and MSFT using their latest public filings. For each company, report the statement period, filing publication date, revenue and operating cash flow with units and source links. Explain missing or non-comparable fields. Research only: do not create or backtest strategies, change portfolios, submit orders, or connect a broker.",
  "max_iterations": 10,
  "include_market_data": false,
  "continue_existing": false,
  "skip_planning": false,
  "cooldown_minutes": 1440
}

Verify the saved state before enabling anything

Copy portfolioId from creation into the get_portfolio argument portfolio_id. Confirm type paper, isActive false, initialValue 0 and one LaunchAgent strategy. Then call list_scheduled_agents with that same portfolio_id and filter disabled. Confirm the intended prompt and schedule. strategyId identifies the scheduled rule; it is a different ID from the portfolio.

Open Agents to inspect the saved schedule. The demonstration ends with it disabled. When enabling your own schedule, review its scope, model/credit settings and cadence first. Reading this page or copying its inactive request does not start the agent.

{
  "portfolio_id": "YOUR_RETURNED_PORTFOLIO_ID"
}

Choose the research scope

Open your saved watchlist and check its members and research purpose. In Watchlist notifications, set AI research brief to Weekly on Friday, choose Save preferences and confirm Alert settings saved. A saved watchlist brief uses those preferences. The separate scheduled-agent example above instead puts a fixed research scope in initial_message and its cadence in schedule.

Write the research request

Ask Aurora to explain changed members, cite public disclosures, distinguish event dates from public availability and identify uncertainty. Limit the request to research unless you separately intend an account action.

Copy the research prompt below after replacing the watchlist description with your intended scope.

Review my selected watchlist. Explain additions and removals since the last review, link supporting public disclosures, and distinguish new public filings from new trades. Report publication dates separately from event dates. Research only: do not change portfolios, run tests, submit orders or connect brokers.

Review before activation

Choose a weekly schedule and time in the app. Review an existing scheduled agent before enabling or replacing it. Programmatic setup supports append/upsert mode and an active setting; specify the intended strategy and portfolio IDs.

Sign in to activate the scheduled review. Reading this guide does not create an agent, start a run or change a deployment.

Continue exploring