APC and APX, two components of a portable agent project, separate portable context from runtime availability. APC stores project contracts in the repository, while APX detects local runtimes via `apx …
Open Interpreter released a fork of Codex that emulates Claude Code and Kimi's harness for cheap models, while OfficeCLI introduced a single binary enabling AI agents to read, edit, and render Word/Ex…
Galda, a new interface layer for Claude Code and Codex, lets developers manage tasks and review AI-generated summaries without digging through Git, supporting dark mode and preserving existing skills …
A software engineer argues that big tech companies, including Meta, are misusing AI by creating internal token-usage leaderboards that incentivize wasteful AI interactions, and that the industry must …
A Hacker News user questioned the necessity of 'skills' in AI agent frameworks like Claude Code and Codex, arguing that well-organized Markdown docs referenced by an AGENTS.md file could serve the sam…
Herdr, a new terminal tool, positions itself as 'to coding agents what tmux is to terminals,' addressing the gap between classic multiplexers and agent-aware orchestration. The tool provides a control…
Vercel's AI Gateway now supports Qwen 3.8 Max, a 2.4-trillion-parameter model handling text and vision tasks with a 1-million-token context window. The model, accessed via model ID alibaba/qwen3.8-max…
An agent skill is a directory containing a skill.markdown file and supporting materials that AI systems like Claude, ChatGPT, and Codex load in stages to complete specific tasks, according to a techni…
Surface, an experimental KDL-based specification format for describing applications without prescribing implementation, has been released as version 0.1 by developer etoxin. The format allows humans a…
OpenAI's Codex users are losing banked rate-limit resets that expire 30 days after being granted, with no warning, as complaints mount in GitHub issue openai/codex#28888 and on the OpenAI Developer Co…
A developer argues that AI startups fail together because model providers absorb their capabilities, citing Jasper's decline after ChatGPT and Windsurf's collapse when Anthropic cut off API access. Th…
A developer released Episko, an open-source (MIT) Rust tool for organizing Claude Code agents, offering project overviews, integrated terminals, session history, and cost tracking. The tool, available…
Anthropic's Claude Code maintains a significant lead in the AI coding-agent sector despite rivals like OpenAI's Codex and open-source models offering cost savings, according to a report from Vera. The…
A former Google employee who joined OpenAI six months ago describes a culture where code wins over design docs and meetings, with weeks passing with barely a meeting. The author notes that OpenAI's in…
Tadashi Shigeoka released omh-issue-loop, a Hermes Agent skill that automates the full journey from a GitHub issue to a pull request by orchestrating Codex CLI for implementation and Claude Code CLI f…
JetBrains released the Kotlin Benchmark, an official benchmark for evaluating AI coding agents on Kotlin software engineering tasks, with 105 tasks from open-source repositories. In the first results,…
Wienerdog, an open-source tool under MIT license, provides persistent memory and self-improving skills for AI coding agents Claude Code and Codex, requiring only an existing Anthropic or OpenAI subscr…
Supabase open-sourced Evals, a free benchmark that scores AI coding agents on real backend tasks, on August 1, 2026 under an Apache-2.0 license. Early results show that with access to Supabase's 'skil…
A developer's analysis shows that the same DeepSeek V4 Flash model produces different long-task outcomes depending on the agent runtime, such as Codex versus Claude Code. The developer argues that the…
Paolo Anzini released MicroCodex, an ultra-lightweight coding agent written in C++23 that runs locally in the terminal, with a binary under 1MB and the codebase comprising 8,000 lines. The tool suppor…