●VERSION — v2.1.227 landed on August 10 with no new features, just fixes around plan detection and CI behavior●AUTO — Two days remain until August 14, when auto mode becomes the default in Claude Code for Pro, Max, and Team●CI — Bash commands no longer fail across the board under claude-code-action with allowed_non_write_users on GitHub-hosted runners●BILLING — Sessions started with an expired token could misread your plan and nudge Max users toward usage credits; that is now fixed●SUNSET — The legacy Workbench and the experimental prompt tool APIs retire on August 17, five days out●PRICE — Sonnet 5 promo pricing at $2/$10 per Mtok runs through August 31, moving to $3/$15 on September 1●VERSION — v2.1.227 landed on August 10 with no new features, just fixes around plan detection and CI behavior●AUTO — Two days remain until August 14, when auto mode becomes the default in Claude Code for Pro, Max, and Team●CI — Bash commands no longer fail across the board under claude-code-action with allowed_non_write_users on GitHub-hosted runners●BILLING — Sessions started with an expired token could misread your plan and nudge Max users toward usage credits; that is now fixed●SUNSET — The legacy Workbench and the experimental prompt tool APIs retire on August 17, five days out●PRICE — Sonnet 5 promo pricing at $2/$10 per Mtok runs through August 31, moving to $3/$15 on September 1

TAG

retry design

1 articles

← Back to all tags

Claude API¹ self-repair¹ error handling¹ production¹

⬡ API & SDK/2026-06-27Advanced

Designing the Give-Up Condition in Self-Repair Loops: Four Error Classes, Four Retry Budgets

LLM self-repair loops break on the fantasy that 'if you keep fixing, it eventually passes.' Classify errors into four classes, give each its own retry budget. Working TypeScript and real cost numbers included.