GitHub (anthropics/claude-code) · Oct 28, 2026 · 3:22 AM

Claude Code 2.1.312 stops overwriting big CLAUDE.md files it only partly read, and caps hooks at 24

Released Oct 27, Write now makes Claude read a very large CLAUDE.md, rules or @-included file in full (or edit in place) instead of clobbering it after seeing only the start, and prompt and agent hooks run at most 24 at a time per session, subagents included. On Bedrock, Vertex, Foundry and Mantle, unrecognized model names like custom deployments now get the shorter system prompt; set CLAUDE_CODE_SIMPLE_SYSTEM_PROMPT=0 to keep the full one.

Read at github.com

Claude Code

An AI coding agent for terminal, IDE, web, and Slack

More news

View all

Intent 2.231.2 adds Codex and Claude Code provider profiles that fail over when one hits quota

Augment's agent workspace now lets you run Codex or Claude Code under named profiles, each with its own env vars (say, a separate work account), and mark some as backups so Intent switches to the next one when a profile hits quota, including Claude Code's weekly and session limits. Queued PRs also show their GitHub merge queue position and ETA…

Intent Updates 57 minutes ago

Fortinet's FortiGate firewalls can now inspect LLM prompts and MCP tool calls inline

Announced Oct 26, the cloud delivered FortiGuard Inline AI Security Service scans prompts, model responses and MCP tool lists, calls and responses in both directions, redacting 35 types of PII and enforcing per team model permissions, budgets and rate limits. If your company runs FortiGate, expect your agents' MCP traffic to start hitting policy checks you didn't write…

Fortinet 3 hours ago

Claude Code adds an env var that lets skip-permissions mode run rm -rf ~ without asking

Since v2.1.311, setting CLAUDE CODE EVEN MORE DANGEROUSLY SKIP PERMISSIONS=1 lifts the short blocklist (like rm rf ) that dangerously skip permissions still prompts for, and Anthropic's Daisy Hollman confirmed it's not a joke. Managed policy settings still apply on top, so if you run unattended agents, keep limits enforced outside the prompt…

X (Daisy Hollman, Anthropic) 4 hours ago

Cloudflare: an SGLang cache bug let GLM-5.3 on Workers AI leak other customers' data in 155 responses

Disclosed Oct 26, prompts over 681,280 tokens to @cf/zai org/glm 5.3 between Aug 27 and Sep 9 hit an SGLang disaggregated prefill/decode bug that left stale KV cache pages in place, so replies could include text from other tenants' requests. Cloudflare says customers need do nothing, but anyone self hosting SGLang in disaggregated mode should upgrade to v0.5.22 or later…

Cloudflare Blog 5 hours ago

Codex CLI 0.165 adds a built-in Vertex AI provider and turns on memory v2 for fresh profiles

Released Oct 27, you can now route Codex through Google Cloud with model provider = "gcp" , GPT 6.1 Terra joins the model catalog, and fresh profiles default to memory v2, which can import other agents' memories. It also closes a marketplace URL check gap that could let Git connect to a host other than the one allowed, so it's worth updating…

GitHub (openai/codex) 6 hours ago

Spotted something we missed? Start a thread.