shell: context pill tooltip states the real tiers, not the archivist
The tooltip claimed 'Archivist begins compacting around 70%'. The archivist does not compact. It reads journal/ and writes system/synthesized/ - a different substrate from the conversation entirely (core/archivist/mod.rs:5). The real advisory tiers are 80/90/95 in ConsciousnessEngine:: on_response, and they never force compaction. Icon thresholds moved from 85/70 to 90/80 to match.
This commit is contained in:
parent
57e53adac3
commit
525157250a
1 changed files with 6 additions and 3 deletions
|
|
@ -398,14 +398,17 @@ Inline w/ backslash and round brackets \\(e^{i\\pi} + 1 = 0\\)
|
|||
// it shows both numbers and the ratio between them,
|
||||
// which is the only reading that means anything.
|
||||
visible: Ai.context.known || Ai.tokenCount.total > 0
|
||||
icon: Ai.context.percent >= 85 ? "error"
|
||||
: Ai.context.percent >= 70 ? "warning"
|
||||
// Tiers mirror ConsciousnessEngine::on_response exactly:
|
||||
// 80 / 90 / 95 are the three compaction-warning tiers.
|
||||
// They are advisory — the engine never forces compaction.
|
||||
icon: Ai.context.percent >= 90 ? "error"
|
||||
: Ai.context.percent >= 80 ? "warning"
|
||||
: "token"
|
||||
statusText: Ai.context.known
|
||||
? `${Math.round(Ai.context.used / 1000)}k/${Math.round(Ai.context.limit / 1000)}k · ${Ai.context.percent}%`
|
||||
: (Ai.tokenCount.total > 0 ? `${Ai.tokenCount.total}` : "—")
|
||||
description: Ai.context.known
|
||||
? Translation.tr("Context: %1 of %2 tokens (%3%)\nCeiling comes from this agent's configured context window.\nArchivist begins compacting around 70%.")
|
||||
? Translation.tr("Context: %1 of %2 tokens (%3%)\nCeiling is this agent's configured context window.\nAdvisory warnings at 80/90/95% — compaction is never forced.")
|
||||
.arg(Ai.context.used.toLocaleString())
|
||||
.arg(Ai.context.limit.toLocaleString())
|
||||
.arg(Ai.context.percent)
|
||||
|
|
|
|||
Loading…
Reference in a new issue