CLAUDE LABJP
CHROME — Compliance API session retrieval now covers Claude in Chrome transcripts as well. Enterprise beta, added on September 182.1.278 — Claude Code has not moved past 2.1.278 since September 19. That release made the auto-mode classifier server-side by default, and the classifier itself is not billed11/17 — The undocumented client-certificate fallback setting stops being accepted on November 17, fifty-six days out. An in-app warning starts on November 3TMPDIR — Windows shells do not set TMPDIR. Commands written out of habit with $TMPDIR fail with Permission denied, and the failures are quietly retriedNEW — The morning a deprecation notice appeared: checking the version before rewriting the configTOOLS — How far should approvals be automated? Lining up Claude Code hooks and permissions next to Antigravity's Deny / Ask / Allow makes your own boundary easier to drawCHROME — Compliance API session retrieval now covers Claude in Chrome transcripts as well. Enterprise beta, added on September 182.1.278 — Claude Code has not moved past 2.1.278 since September 19. That release made the auto-mode classifier server-side by default, and the classifier itself is not billed11/17 — The undocumented client-certificate fallback setting stops being accepted on November 17, fifty-six days out. An in-app warning starts on November 3TMPDIR — Windows shells do not set TMPDIR. Commands written out of habit with $TMPDIR fail with Permission denied, and the failures are quietly retriedNEW — The morning a deprecation notice appeared: checking the version before rewriting the configTOOLS — How far should approvals be automated? Lining up Claude Code hooks and permissions next to Antigravity's Deny / Ask / Allow makes your own boundary easier to draw
TAG

Scheduled Runs

1 articles
Back to all tags
Related:
Claude Code1Automation1Error Handling1Operations1
Claude Code/2026-08-03Advanced

Existence Checks Pass, Writes Fail — Probing Capabilities Before an Unattended Run

A directory existing and a directory being writable are two different facts. Measured results from five broken-environment cases, why static clues cannot predict the worst of them, and a capability-probe preflight for unattended Claude Code runs.