Your first investigation

A good first investigation is narrow enough to finish, but open enough to reveal something you did not already know. Start with a subject and a time window; add a place or behavior when the archive search would otherwise be broad.

1. Ask for the observation first

Begin with what the data can directly establish. Aircraft positions, timestamps, registrations, callsigns, advisories, and public records are observations. Intent is usually an inference.

Show the flights by N12345 within 30 nautical miles of Edwards AFB from May 1 through May 31. Group them into visits and summarize the duration of each visit.

This is stronger than asking “Why was N12345 spying on Edwards?” The first question produces evidence you can evaluate; the second embeds a conclusion before the search begins.

2. Add context in a second pass

Once the relevant flights are identified, ask HEXBIRD to add only the context that can change the interpretation:

• registration and ownership history;
• callsigns and operator metadata;
• nearby airports, airspace, and infrastructure;
• TFRs or NOTAMs active at the time;
• ACARS or datalink messages around the flight;
• TCAS / ACAS events involving either aircraft.

For those visits, check ownership history and nearby aviation or industrial infrastructure. Separate direct evidence from plausible interpretation.

3. Ask for an artifact

Tell HEXBIRD what form will help you inspect or communicate the result. Useful outputs include a track map, event timeline, comparison table, chart, aircraft dossier, downloadable dataset, or self-contained HTML report.

Build a HEXBIRD artifact with a map with one color per visit, mark the closest approach to the base boundary, and include a table linking each claim to the supporting flight and timestamp.

4. Check coverage and uncertainty

Absence from an ADS-B or datalink dataset does not prove an event did not occur. Before treating a result as exhaustive, ask:

• Which sources and tables were searched?
• What time zones and date boundaries were used?
• Was the result capped or sampled?
• Were there coverage gaps or aircraft with hidden or missing identity fields?
• Which findings are observed facts, and which are judgments?

A complete starter prompt

Investigate N12345 near Edwards AFB from 2026-05-01 through 2026-05-31 UTC.

1. Group its tracks into distinct visits within 30 NM of the field.
2. Summarize each visit's entry time, exit time, duration, minimum distance,
   altitude range, and apparent flight pattern.
3. Check registration history and nearby aviation infrastructure only where
   they help interpret a visit.
4. Produce a map and a compact evidence table.
5. State coverage limits and separate observations from judgments.