Intel / repo

claude-code v2.1.318

Source: github.com
Author: ashwin-antTop Viber
Date: Published Nov 4, 2026

Context

Claude Code 2.1.318 is a long maintenance release, and its clearest thread is making unattended runs fail less silently (release notes). A stored claude.ai sign-in now reports when it ends: `claude auth status` shows `loginExpiresAt`, `claude -p` JSON results carry `credential_expires_at`, and `claude -p` text output prints a one-line warning to stderr starting 10 days before the sign-in ends. When the sign-in does run out, the error now says so and, in `claude -p`, names `claude auth login` and `claude setup-` as the ways back. For anyone running Claude Code from scripts or schedulers, that turns a sudden sign-in failure into something a job can detect ahead of time.

Much of the rest concerns `~/.claude.json`, the configuration file that holds trusted folders and server settings, especially on shared network home directories. Anthropic lists fixes for logout and MCP removal commands that reported success when the file could not be written, a stored login being replaced by an empty file when a network home directory ran out of space, and non-interactive launches overwriting an invalid file with defaults; that start now fails and explains how to restore it. Unattended runs also stop rewriting the file when nothing changed. The matching Agent SDK 0.3.318 release exposes the same expiry date and a new `config_file_invalid` startup failure reason to SDK apps.

Other notable items: `--continue` could lose the earlier conversation after a malformed tool call's retries ran out (a regression from 2.1.314), idle could replace a conversation with a reply that was not a summary, and Claude Code no longer installs or updates the IDE extension when started in an IDE terminal.

Key takeaways · AI-distilled

Background subagents now move a shell command to the background after 3 minutes on all providers, so the prompt cache does not expire while a long command runs.
Non-interactive launches no longer replace an invalid ~/.claude.json with defaults: the start fails, explains how to restore the file, and leaves it as is. Unattended runs also stop rewriting the file when nothing in it changed.
deniedMcpServers URL entries now also block Anthropic-provided servers whose upstream URL matches, and connectors in cloud sessions and on self-hosted runners; any other connector must be denied by name.
Claude Code no longer installs or updates the IDE extension when started in an IDE terminal; the release notes point to /ide from an outside terminal or the editor's extensions view instead.

Terms in this piece · Glossary

token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context compaction — Summarizing an agent's earlier conversation to free room in the context window so a long session can keep going.

Read the source github.com