github Goose (Block) · 1h ago · agents

Goose v1.45.0 adds Opus5 support, tab completion, and structured summaries

from English

The Goose project released version 1.45.0, introducing several new features alongside bug fixes and improvements.

Added support for the latest Gemini models and Opus5 with adaptive thinking.
Introduced tab completion and provider switching for the `model` command.
Enabled configurable GOOSE_DOCS_ROOT for air-gapped documentation access.
Implemented structured summary output with template rendering for compaction.
Upgraded to rmcp 2.0 and disabled the scheduler by default for ACP.

This release enhances model compatibility and provides better configuration options for users operating in restricted environments.

Importance 1/3 Trust 2/3 Goose (Block) Cursor AI agents Code generation

Read original

media TLDR AI · 12d ago

Claude Code adds in-app browser; Cursor builds general agent

Anthropic has added an in-app browser to Claude Code on desktop, allowing the tool to read, click through, and interact with websites and docs similarly to local dev servers. Concurrently, reports indicate that Cursor is developing a general-purpose AI agent designed to handle emails, texts, spreadsheets, and engineering tasks.

lab Claude Code Releases · 19d ago · 2 views

Claude Code v2.1.207 enables Auto mode by default and defaults to Claude Opus 4.8

Claude Code version 2.1.207 makes Auto mode available without opt-in on Bedrock, Vertex AI, and Foundry, while changing the default model for these platforms to Claude Opus 4.8.

arxiv arXiv cs.AI · 43d ago · 1 view

Oracle Signals in Agent-Authored Test Code

An empirical study of 86,156 test-file patches from 33,596 agent-authored PRs reveals that 80.2% of test patches contain weak or no explicit oracle signals. Strong-oracle test files significantly improve merge likelihood (OR = 1.28, p < 0.001) after adjusting for multiple factors, indicating test file presence alone overestimates verification strength.

github AutoGPT · 1h ago Live

AutoGPT platform beta v0.6.70 adds Expert entity and configurable transcription endpoints

The AutoGPT team released version 0.6.70 of the autogpt-platform-beta, introducing new features for CoPilot and agent management alongside several bug fixes.

media TLDR AI · 6h ago

xAI launches Build Mode; researchers urge AI slowdown pact

xAI has launched "Build Mode" for SuperGrok Heavy subscribers, enabling users to generate, edit, preview, and publish websites, apps, games, and dashboards directly from chat without setup. Simultaneously, over a thousand employees from frontier AI companies have signed a statement requesting that the US government support an international effort to develop tools for deliberately pacing the frontier of automated AI development.