docs: sync ARCHITECTURE.md and benchmark pricing to match code#1720
Merged
esengine merged 1 commit intoMay 25, 2026
Merged
Conversation
Three stale-doc fixes: 1. ARCHITECTURE.md §4.3 — Replace deleted /pro single-turn arming with current /model + config model selection. Add history note referencing v0.50.0 removal (esengine#1657, esengine#1630). 2. ARCHITECTURE.md §4.4 — Replace non-existent FAILURE_ESCALATION_ THRESHOLD=3 auto-escalation with the <<<NEEDS_PRO>>> model self- report mechanism. Clarify no failure-counter or scavenge/storm counting exists. 3. benchmarks/real-world-cache/README.md — Fix pricing numbers to match src/telemetry/stats.ts (flash cache-hit was 10x too high, pro numbers were all off). Recalculate both cost tables and savings percentages accordingly.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three stale-doc fixes where documentation no longer matches the actual codebase:
1. ARCHITECTURE.md §4.3 —
/prosingle-turn arming (removed in 0.50.0)Rewrote to describe the current
/model flash|procommand and config-based model selection. Added a history note referencing #1657, #1630.2. ARCHITECTURE.md §4.4 —
FAILURE_ESCALATION_THRESHOLD=3auto-escalation (never existed in code)Replaced with the actual
<<<NEEDS_PRO>>>model self-report mechanism. No failure-counter, no scavenge/storm counting — purely LLM-initiated. The"failure-threshold"variant inEscalatedEvent.reasonis also dead code; kept the type union for now.3.
benchmarks/real-world-cache/README.md— pricing numbers out of sync withsrc/telemetry/stats.tsThe displayed pricing for v4-flash cache-hit was 10x the code value (0.028 → 0.0028). v4-pro pricing was entirely wrong (0.139/1.667/3.333 → 0.003625/0.435/0.87). Recalculated both cost tables and savings percentages. The headline cache-hit ratio (99.82%) is unchanged.
Verification
grep FAILURE_ESCALATION_THRESHOLD src/ -r→ zero matches (only in docs)grep "/pro" CHANGELOG.md→ confirms removal in 0.50.0 (refactor(config): drop preset abstraction, expose model + effort directly #1657, feat(presets): drop auto + /pro — flash/pro are the only knobs now #1630)stats.tspricing cross-referenced with case study cost math