Using Agent-Reach to read content across platforms, how to avoid treating old messages as new ones?

Filter recent messages; first, agree on which time format to use?

Clearly define the list's purpose: newly discovered messages are sorted by first collection time, while viewing content published on a specific day filters by original publication time. Each list can use distinct labels so readers know which type of timeline they are viewing. Preserve the meaning of these times in collection results, continuing the same convention when generating summaries or history pages later.

How to handle missing dates or future time references?

It is recommended to set readable statuses for time information, distinguishing between sources providing full timestamps, those providing only dates, and those where timing is temporarily undetermined. Materials with undetermined publication times can still enter the newly discovered list, but the page should display the actual ingestion time and time status, avoiding fabricating a precise-looking publication time.

When a discovery time is later than the current time, first record the original value and discovery time, then check for timezone interpretation or input format issues. Before processing, leave it in a pending state to avoid entering the daily ranking list based on publication time. This describes our own data processing method; we cannot infer that Agent-Reach already possesses corresponding correction or anomaly detection capabilities.

How to preserve records when updating entries for duplicate content?

You can initially use existing content identifiers or links to locate records, retaining the first collection time, and separately recording the current discovery time. When the same content appears again, check if summaries or attached information have changed substantively before deciding whether to display it as an update. This allows clearer explanation of list order and content changes during continuous collection tasks.

Under what conditions can you increase platform usage and operational frequency?

First check per-round discovery counts, time completeness, and duplicate records using a small range of materials, confirming that failure sources have explicit status flags. When a platform returns no content, record whether it is due to lack of new material, read failure, or temporary inability to distinguish, facilitating subsequent judgment of coverage. Once stable, add more sources and continue saving brief run results for each round.

Check results with a fixed input

Use three prepared records to check date handling: A was published at 2026-10-20T23:30:00+08:00, B has a date only, and C has no publication time. Give all three an October 21 collection time, then filter October 20 in Beijing time. A belongs in the publication-day list, B needs a precision label, and C belongs only in the newly discovered list. Reading A again must preserve its first collection time. This is an acceptance example, not a platform collection test.