Full-text search across 58891 articles. Combine with topic and date filters; results sorted by relevance.
The KwaiKAT Team at Kuaishou released KAT-Coder-V2.5, an agentic coding model trained on over 100,000 verifiable repository environments spanning 12 languages. The model is available through StreamLake, with an open-weig…
Lawrence Wong, writing under the pen name Ahlimosa, identifies context isolation and task coordination as the biggest issues facing AI coding agents in multi-project environments. He argues that while isolated agents mai…
Anthropic's Claude Code agent autonomously redesigned a UI page, executed local file changes, staged them, and pushed a commit to a remote Git repository, completing the task in roughly 4 minutes compared to 20 minutes f…
Cursor launched Router on July 22, an automatic model selector for Teams and Enterprise customers that classifies each coding request before a model runs, delivering frontier-quality performance at 60% cost savings. Curs…
A developer measured the token overhead of MCP servers in Claude Code sessions, finding that the official GitHub MCP server adds ~62,000 tokens per 20-turn session—costing about $0.19 in tool-definition overhead alone. T…
Anthropic's July 17 Claude Code update changes /fork to create an independent background session instead of an in-session subagent, introduces /subtask as its replacement for isolated in-session tasks, and adds default c…
A developer argues that while AI coding agents are heavily tooled with linters, CI, and memory stores, the human operator's own workflow remains largely unharnessed, relying on markdown files and honor systems. The post …
A developer built a custom three-line status line for Claude Code that displays plan quota, context usage, session cost, and automation health, driven by JSON piped from stdin. The script caches the JSON to a temp file s…
Google's Gemini 3.5 Pro is months behind schedule as the company struggles to improve its coding capabilities, Bloomberg reported, citing ten current and former employees. Google said in May it expected to ship the model…
OpenAI on July 15 launched its first branded hardware product, the $230 Codex Micro macro pad, co-built with mechanical keyboard maker Work Louder, designed to give developers a physical command center for managing AI co…
GeekyAnts built a bridge from AI-generated React code to editable Figma files by reading React Fiber instead of the DOM, solving problems like poor Auto Layout support and missing reusable components. The company found t…
Deja, a zero-dependency open-source binary, turns local agent conversation histories from Claude Code, Codex, and opencode into a searchable memory layer that works over SSH. The tool indexes gigabytes of logs retroactiv…
GitHub shipped CodeQL 2.26.0 on July 10 with a new static analysis query, js/system-prompt-injection, that detects untrusted user input flowing into AI model system prompts in JavaScript and TypeScript, flagging vulnerab…
A developer details how to properly deploy Claude Code on AWS Bedrock, covering IAM policies, SCPs, and a governance model that most enterprise teams get wrong. The guide highlights common pitfalls like model ID format m…
PlanWright, a control plane for AI coding agents, launches to solve the bottleneck of human-paced planning and acceptance by inverting those ceremonies. The tool synthesizes raw input like transcripts and Slack into stru…
Aftr, an open-source Node.js controller by Arman Luthra, lets developers drive Adobe After Effects programmatically over WebSocket or MCP, enabling autonomous, spec-driven video pipelines that self-correct by rendering, …
Anthropic shipped Claude Code 2.1.207, making auto mode the default on Amazon Bedrock, Google Vertex AI, and Microsoft Azure Foundry. The auto-mode classifier, a second model that reviews pending actions against conversa…
A developer built an unattended autopilot that uses Claude Code to improve its own environment daily, with cost and repetition safeguards. The system runs at 5 AM via launchd, picking one improvement task per session, an…
Microsoft released VS Code 1.128 on July 8, introducing multi-chat agent sessions that allow parallel AI conversations within a single workspace, and making Copilot Vision generally available without requiring IT admin a…
Claude Code's system prompt includes unnecessary tool definitions and instructions that inflate token usage on every request. Users can reduce payload size by disabling bundled skills, removing individual tools via deny …