DeepSeek released DeepSeek Harness, an agentic coding framework in developer preview alongside the DeepSeek V4 Pro model, which uses a plugin architecture and a local web app interface. Installation r…
Boris Cherny, an AI engineer, outlines practical loop engineering for AI agents in Claude Code and Codex, describing how he runs five to ten agents in parallel and relies on built-in primitives like g…
Pipe, a language-embedded agent sandbox, distinguishes itself by placing sandbox control inside the programming language itself, with ratchet semantics that prevent privilege escalation, structural ga…
HolyClaude, an open-source tool for Claude Code, reached 2.4k stars on GitHub, and its developer is now hosting it as a managed service on Fly.io, giving each user a dedicated Linux VM with pre-instal…
Anchorstate Lab released GMR, a lightweight CLI runtime that binds manual judgments and notes to recomputable observations in a repository, replaying the observation that generated them and warning wh…
DeepSeek released Harness v0.1, an open-source, MIT-licensed agent runtime, in developer preview on August 16, positioning it as a direct competitor to Claude Code. The same day, DeepSeek quadrupled V…
A developer proposes a tiered context strategy for coding agents to reduce token costs, suggesting a small always-on index file, path-scoped detail files, and on-demand injection. The approach caps th…
Anthropic's Claude Code, an AI coding tool that operates inside a terminal, requires a loop of prompting, checking, and approving changes rather than a single prompt, according to a user guide. The gu…
Leo Proger released Claude First Customer Finder, a Claude Code skill that turns a startup URL or product idea into a qualified shortlist of potential first customers using public pain, demand, and ti…
Reachpad now includes free access to Gemma 4 31B and Gemini 3.6 Flash coding models on every pad, usable through OpenCode without an API key. The models are included with the pad, with a daily allowan…
A developer has released 'Caged Code', a webpage that embeds the vanilla Claude Code binary and connects it to a notebook environment via WebSocket, allowing users to run Claude Code through their Ant…
A developer has documented a workaround for running Kimi Code's K3 model inside Anthropic's Claude Code terminal tool, addressing configuration pitfalls including a misleading model ID and authenticat…
DeepSeek released Harness, an MIT-licensed agent framework built on the Cordis plugin kernel, into developer preview. The framework treats every agent capability as a plugin, enabling model, tool, and…
Mirage, the AI startup behind the video-editing app Captions, launched a 24-hour AI-generated news channel called Mirage News Network, which reached 50,000 viewers on X and drew 824,000 impressions on…
RudVlad473 released iluvatar, a Claude Code skill that classifies an app idea into computer-science terms before any technology is named, emitting a schema-validated HIGH-LEVEL-ARCHITECTURE.md file th…
Anthropic investors are betting on a $2 trillion valuation for the AI company's potential IPO, reflecting confidence in its enterprise adoption and agentic AI shift. The valuation target hinges on Ant…
Microsoft is shutting down Copilot features including Deep Research, AI Podcasts, Group Chats, Copilot Labs, and the Mico mascot on August 18, and merging its consumer Copilot app with Microsoft 365 C…
Cadre.rocks has launched a local-first board for orchestrating coding agents in Git worktrees, enabling parallel agent runs with human approval gates. The tool supports Claude Code by Anthropic, Codex…
Anthropic announced that auto mode is now the default permission mode for new Claude Code sessions on Pro, Max, and Team plans as of August 14, 2026. The mode routes actions through a separate classif…
A developer benchmarked a homelab agent memory stack on the LoCoMo benchmark and improved accuracy from 63% to 80% by adding hybrid search with BM25 and a local reranker. The developer found that dens…