An open-source model system called Fusion-MOA, built by a developer team, achieves a 50% pass rate on real-world terminal engineering tasks using a 27B-parameter model with consultations from other mo…
Local AI Weekly #1 reports that Moonshot AI released Kimi K3 on July 16, a 2.8-trillion-parameter Mixture-of-Experts model with a 1M token context window under a Modified MIT license, making it the la…
Oqoqo, a new tool launched on Hacker News, enables developers to build realistic evals and custom benchmarks for real-world tasks, measuring agent-friendly product surfaces against Codex, Claude Code,…
Revise, a new AI-native word processor, has launched first-class Model Context Protocol (MCP) support, making it the world's first DOCX-compatible editor that is agent-agnostic and composable with any…
The ZWISERFIT engineering team reported that its agent orchestration system, LAO, detected, repaired, and permanently prevented five distinct failures across three AI agents in a 24-hour autonomous cy…
Hermes plugin developer released an adversarial LLM reversal tool, hermes-agent, that catches large language models sandbagging in real time by forking two disposable clones every seventh turn to inte…
Computer use agents, which let large language models operate a full computer through screenshots, clicks, and keystrokes, remain challenging to build, as the OSWorld benchmark shows. Introduced by Ant…
DeepSight, an open-source tool from Reality-Shifting-Tech, lets text-only LLMs like DeepSeek see and interact with the real world using zero tokens and on-device vision, supporting macOS via Apple Vis…
Ocean, a new tool from Mosaic Inc., centralizes AI agent sessions from Claude Code, Codex, OpenCode, Hermes, Pi, and other harnesses into a searchable platform, enabling teams to recover context behin…
A new technical post by an unnamed author details how to build a production-grade agentic harness for LLM-based agents, upgrading a basic loop with typed tools, a plan DAG, tiered memory, verification…
Tencent's open-source AI agent memory system, TencentDB Agent Memory, reached #1 on GitHub Trending this week, adding nearly 2,000 stars in a single day to reach 14,600 total. The tool provides persis…
Amazon Web Services Inc. launched Kiro Crew, an autonomous agentic orchestrator that keeps AI coding agents running 24/7, allowing developers to start tasks and walk away. Built on the Kiro developmen…
Ante, the agent harness developed by the company behind this test, passed all 10 Terminal-Bench 2.1 tasks using the DeepSeek DeepSeek-V4-Flash-0731 model, while Ante-short passed 9 and Pi 0.73.1, Open…
Mert, creator of MoneyCo (thecompany.money), launched a coding agent built for accounting that connects to multiple QuickBooks and Xero accounts via API instead of MCP, deploys client files in a sandb…
Palo Alto Networks' Unit 42 revealed that a threat actor in Zhuhai used DeepSeek, the open-source Hermes agent framework, and Telegram to orchestrate over 460 attacks on internet-facing systems. The a…
AGENT EVE, a persistent online world where autonomous AI agents build, trade, ally, and betray, is now live at agenteve.io, with daily settlements called the Reckoning. The game is designed for agents…
Hunt.io and security researcher Bob Diachenko discovered an exposed server in Hong Kong that revealed an active intrusion attempt against Thailand's Ministry of Finance between July 9 and July 13, 202…
Y Combinator has open-sourced QM, an AI agent harness it built to manage fleets of internal agents, under the MIT license on GitHub. The harness, designed for Slack and the web, provides scoped memory…
Tilde, a harness SDK platform for building and self-hosting AI agents, released a code review agent example on GitHub. The project decomposes features from OpenClaw and Hermes into cloud API building …
A security researcher detailed an autonomous AI attack that exploited exposed Langflow, n8n, and Marimo instances but was foiled when the agent left a public web server broadcasting its own tool calls…