ADE Index is a sourced, ranked directory indexing 975 agentic coding tools, with entries organized by category such as desktop apps. The featured listing, Paseo, is an Apache-2.0 desktop and mobile app for orchestrating multiple parallel coding agents with worktree isolation and agent teams.
Agentic Coding
智能体编程
Workflows where agents autonomously analyze code, benchmark and validate changes, used here to support performance optimization.
Curated Intelligence
Uber Eats rebuilt major parts of its search pipeline, shifting the optimization target from backend API response time to above-the-fold render completion. Server-side cached pagination, async rendering, removal of low-value retrieval strategies, product-level embeddings and a redesigned ads path cut end-to-end latency by 50% and above-the-fold latency by over 200ms. The team also applied agentic coding workflows to identify and validate further optimizations, and is exploring zero-pass ranking and HTTP multipart streaming.
Investor Yuchen Jin argues that development is shifting from writing code and then testing it to building the verifier first and letting agents generate code until it passes. The framing places automated correctness checks at the center of agentic coding workflows.
AWS has open-sourced aws-emulator, a local emulator for AWS services that frees developers from depending on live cloud environments. Much of the code was written with heavy agent involvement, and the full commit history plus every agent-filed issue are public, including a 'things-that-do-not-work' folder tracking behaviors that still diverge from real AWS. It offers a concrete, observable case study of what engineers do once agents write the code.
Early hands-on tests of Fable 5.2 show it beating Opus 5.5 across agentic coding, software engineering and computer-use benchmarks, with a roughly 45.5% coding lead, 69.9% on software engineering, a 1M-token context window, up to 128K output tokens, up to 30% faster output and a 20% lower API price. Users are already generating 3D worlds, games and interactive scenes at scale, though token burn and cost at high reasoning settings are the main complaints.
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, claiming notable gains across coding, reasoning, scientific research, cybersecurity and robotics, with up to 30% faster output and 20% lower cost. Fable 5.2 scored 49.1% on the agentic coding benchmark FrontierCode 2.0, beating GPT-6 Astra at 42.8% and Gemini 4 Carbon at 41.1%, and also led on multidisciplinary reasoning tests. Anthropic demonstrated it coordinating over 1,000 agents over eight days to build a bootable operating system from scratch.
CodeRabbit has built an AI code governance platform on OpenAI's Codex and Daybreak to review, validate and secure agent-generated code. The company claims a 20x increase in software shipped with the same engineering team, serving over 20,000 organizations and 150,000 open-source projects, and has launched CodeRabbit Secure.
Karpathy used Claude Opus 5.5 to mine 43 million Hacker News comments spanning 20 years, distilling 6,454 repeatedly recommended videos into the login-free HackerTube over a weekend. He then one-shotted TristanTube with Claude Code in three minutes, showing how coding agents can compress weeks of engineering into hours.
AI coding tools are multiplying pushes and automated jobs, and GitHub's repeated outages have exposed the limits of conventional code hosting. OpenAI has disclosed its internal Git backend Pando, reporting a nearly 80% drop in median push latency, and plans to acquire code-hosting tool Pierre; Anthropic has moved its main repository onto a self-built system, and Cursor now lets developers store projects directly. Code hosting is shifting from general-purpose platforms to custom infrastructure built for AI workloads.
AI chip design firm Ricursive Intelligence has released Ripple-65B Preview, an Apache 2.0 open model pretrained on 12 trillion tokens using a dense 65B Transformer rather than a mixture-of-experts design. It scores 80.52 on TerminalBench 2.1, the best among open US models, though it trails the newly released Inkling 2 Small on other benchmarks; Ricursive puts total R&D and training cost at just $32 million.