luna/oh-my-opencode
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

fix(context-window-monitor): clamp displayed context status percentages so the directive stays trustworthy (fixes #3655)

Browse Source 
Root cause: the context-window-monitor hook computes actualUsagePercentage = (input + cache.read) / actualLimit and renders both 'X% used' and '(1 - X) * 100% remaining' inside a [SYSTEM DIRECTIVE: OH-MY-OPENCODE - CONTEXT WINDOW MONITOR] block that is appended to bash tool output. When resolveActualContextLimit() underestimates the model's real context window (for example a 1M-context Anthropic model that falls back to the 200K default per #3450), totalInputTokens > actualLimit and the rendered numbers go nonsensical (issue #3655 reproduces 144.7% used / -44.7% remaining at 289,370 / 200,000 tokens). Safety-tuned models recognize the >100% / negative-remaining pattern as a tell-tale prompt injection and refuse to follow the directive.

Fix: clamp actualUsagePercentage to [0, 1] before formatting. The 70% threshold check still uses the raw value so the block continues to fire above threshold, and resolveActualContextLimit() is left untouched (the deeper resolver concern is tracked separately as #3450). When totalInputTokens exceeds actualLimit the displayed numbers now read '100.0% used / 0.0% remaining' instead of the impossible >100% / negative pair, and safety-tuned models stop flagging the block as an injection attempt.

Verification: added a regression test (input 289,370, limit 200,000) that asserts usedPct in [0,100] and remainingPct in [0,100]. Test fails before the fix (Received: 144.7) and passes after. Full context-window-monitor.test.ts and context-window-monitor.model-context-limits.test.ts: 15 pass / 0 fail. Typecheck clean.
This commit is contained in:
MoerAI
2026-04-28 16:51:56 +09:00
parent 938b609a91
commit 07064a96f5
2 changed files with 65 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/hooks/context-window-monitor.ts
+9 -2
View File
@@ -65,8 +65,15 @@ export function createContextWindowMonitorHook(
remindedSessions.add(sessionID)
const usedPct = (actualUsagePercentage * 100).toFixed(1)
const remainingPct = ((1 - actualUsagePercentage) * 100).toFixed(1)
// Clamp the displayed percentages so the block stays trustworthy when the
// resolved actualLimit underestimates the model's real context window
// (e.g. a 1M-context Anthropic model that falls back to the 200K default).
// Without clamping, the block would advertise >100% used and a negative
// "remaining" - safety-tuned models flag exactly that pattern as a prompt
// injection and refuse to follow the directive (issue #3655).
const clampedPercentage = Math.min(Math.max(actualUsagePercentage, 0), 1)
const usedPct = (clampedPercentage * 100).toFixed(1)
const remainingPct = ((1 - clampedPercentage) * 100).toFixed(1)
const usedTokens = totalInputTokens.toLocaleString()
const limitTokens = actualLimit.toLocaleString()