lab Claude Code Releases · 22h ago · agents

v2.1.276 fixes 400 error on proxy when ANT

from English

Version 2.1.276 resolves a regression introduced in 2.1.275 where every request failed with a 400 error containing the input tag 'advisor_20260301'. This issue occurred specifically when the ANTHROPIC_BASE_URL environment variable pointed at a proxy or gateway.

Fixed requests failing with 400 error and input tag 'advisor_20260301' on proxies.

Importance 1/3 Trust 2/3 Claude Code Releases Anthropic

Read original

lab Anthropic News · 4h ago

Anthropic partners with Accenture on embedded AI evaluation

Anthropic is partnering with Accenture’s specialist AI business to conduct independent evaluation and red-teaming of its frontier models. This initiative supports Anthropic's commitment to embed evaluators within the company, allowing them to monitor model development and verify safety commitments.

lab Claude Code Releases · 6h ago · 3 views

Claude Code v2.1.277 adds AGENTS.md support and fixes numerous bugs

Anthropic released Claude Code v2.1.277, introducing support for the AGENTS.md file as a fallback for project instructions and adding several configuration options for the Claude apps gateway.

media MarkTechPost · 23h ago · 4 views

Anthropic launches Claude Code Projects in beta with parallel cloud sessions

Anthropic has redesigned Claude Code Projects, shifting from a folder-based structure to a single ongoing conversation where Claude acts as a coordinator. This new architecture allows the system to split work into parallel cloud sessions that continue running even after the user closes their laptop.

lab Claude Code Releases · 1d ago · 7 views

Claude Code v2.1.275 adds send-now key, syncs skills to terminal, and fixes resume crashes

Claude Code version 2.1.275 introduces a new 'send-now' keybinding that interrupts the current turn to queue messages, and synchronizes skills and plugins from claude.ai accounts to terminal sessions. The update also resolves critical stability issues where malformed transcripts or memory entries caused crashes during session resumption.

lab Anthropic News · 1d ago · 13 views

Anthropic launches Life Sciences Verification Program with permissive model access

Anthropic has introduced the beta Life Sciences Verification Program (LSVP), granting verified life science professionals access to its Mythos, Opus, and Sonnet models with safeguards more permissive for biology-related work. The program allows teams to apply for "Standard Use" or "High-risk Use" grants after a verification process reviewing research credentials and security standards.