Dan Luu's analysis of token costs across programming languages reveals how your language choice impacts AI coding agent efficiency, cost, and context usage.
Implant, a VS Code extension by strangepleasures, exposes the editor's internal APIs to external coding agents — a different architectural bet from Cursor or Copilot.
UnYOLO is a credential broker and policy engine for AI agents accessing GitHub. Agents get short-lived proxy tokens instead of real credentials.
us-vs-them tracks human vs AI line-level authorship using diff history — passive provenance for files edited by both humans and AI agents.
"Code was never the hard part" is a lie programmers tell themselves, argues a blog post that hit 117 HN points. Writing code forced understanding AI agents skip.
Gentoo developer Michał Górny took Bugzilla offline after LLM scrapers flooded it from thousands of IPv4 addresses. No pattern to block, no time to fight it.
Revise.io shipped a free DOCX editor with an MCP server that lets AI agents like Claude Code and Cursor edit Word documents programmatically.
Starting August 14, Claude Code will use auto mode as its default permission setting for Pro, Max, and Team users. A separate classifier caught 89% of dangerous commands in testing.
A community-built session-bridge skill lets two Claude Code sessions exchange messages directly, enabling multi-agent coordination without manual context copying.
HAR is an open source project for coordinating multi-agent coding workflows. Early stage, minimal docs, but targets a real gap in AI coding tooling.
Command Code's GOAT plan: $10/month for $70 in credits across 30+ open and closed models, with deals pushing value past $100.
Compass is a local-first knowledge graph engine built in Rust that indexes code artifacts on demand for AI coding agents — no background scanning, no server.
Ocean syncs Claude Code, Codex, and other agent harness sessions into a shared, searchable team workspace — so the reasoning behind code changes doesn't die on someone's laptop.
AWS open-sourced Kiro Crew, a persistent development workspace that lets AI coding agents run multi-step tasks unattended across sessions, tools, and time.
A Hacker News post reveals a developer in an AI-skeptical game-dev community secretly using LLMs to fix bugs in a multi-language toolchain they previously couldn't navigate.
MOTHER is a standalone terminal built for Claude Code with one-click session resume. Listed on Product Hunt, it wraps Claude Code's CLI session management into a visual UI.
A free CLI reads local files from Cursor, Claude Code, Codex, and 15 other AI coding tools, then puts all your token usage in one dashboard with cost estimates and a leaderboard.
mpai adds multiplayer collaboration to existing Codex and Claude Code sessions, letting teams share AI coding context without copy-pasting between terminals.
TamedTable is an open-source LLM harness for data ETL built using AI. The full specification is on GitHub, letting you regenerate or fork the entire codebase.
A new site asks whether your next project can be vibe coded or needs manual work. The question developers now ask before starting.
An AI PM agent evaluated its coding agents and recommended replacing them — a glimpse at where multi-agent orchestration is heading.
A 0-day in Cursor on Windows executes a planted git.exe from any repo root with no user interaction. Mindgard went public with full disclosure after private channels stalled.
An open-source bridge connects Adobe After Effects to AI coding assistants — Claude Code, Cursor, and Google's Antigravity — for motion design via natural language.
A senior UX designer reports PMs shipping AI-generated prototypes to production with design systems slapped on — and the design debt is already cascading.
Redis creator antirez argues developers should shift from controlling code to controlling ideas as AI transforms programming. Key points and takeaways.
CodeMote pairs your iPhone to your dev machine and streams live CLI agent sessions — push notifications when Claude Code or Codex needs approval.
Scritty launched on Product Hunt as a shared, searchable memory system for AI coding agents. Here's what we know about the tool and where it fits.
Herdr is an open-source CLI tool that lets you launch, monitor, and coordinate multiple AI coding agents side by side in a single terminal.
Lore is an open-source tool that captures architectural decisions and team conventions so AI coding agents stop re-asking the same questions.
Jonathan Beard argues AI coding tools handle 80% of the work but leave the hardest 20% — the part that defined engineering skill — to developers who may no longer know how to do it.
A developer vibe-coded a full GPU-accelerated desktop image editor in Rust for Linux — layers, selections, brushes, vector text, and background removal included.
A developer used Claude Opus to autonomously port Slay the Spire 2 to Android with native ARM compilation and touch controls — in about 6 hours of work.
Kibitz lets you run multiple LLMs as 'coworkers' that fact-check each other's output locally, starting with ComfyUI workflows.
KTStack bundles Nginx, PHP, Node.js, Python, databases, and Cloudflare Tunnels into a single native macOS app. It's open source and built as a learning project.
A developer vibe-coded a fruit-slicing game for their MacBook Touch Bar to fill the idle seconds while Claude processes prompts.
A solo builder used Claude for frontend and manual backend work to ship feedbacked.org — a paid feedback marketplace built through constant user iteration.
A new community site lets vibe coders submit projects for review and discovery. Built with vibe coding itself, VibeCoding.is aims to catalog what people are actually shipping.
A vibe coder wants to ditch WordPress entirely and rebuild their site in React or Astro using Claude Code as the primary dev tool. Here's what that actually involves.
A developer used Claude Code to build a custom blockchain where AI agents can create tokens, hold wallets, and execute transactions autonomously via MCP.
A developer built a browser-based animation tool to solve the gap between shipping a vibe-coded app in a weekend and spending weeks on the launch video.
A solo developer scraped luxury hotel prices daily for six months across 2,000+ hotels and built a price-trend site showing when to actually book.
Flama's CLI now lets you spin up a full LLM API with an interactive chatbot UI from a single terminal command. Here's what it does and where it fits.
Papermark's founder publicly accused a developer of passing off forked open-source code as a vibe-coded project, sparking debate on attribution in the AI coding era.
Polygraph gives AI coding agents visibility across multiple repos and persistent session memory, tackling the context fragmentation problem in multi-repo codebases.
VibeSolve is an open-source tool that converts natural language descriptions of optimization problems into Timefold code, outperforming raw LLM prompting for rapid prototyping.
Games disclosing generative AI usage on Steam sell significantly worse than non-AI titles, with penalties reaching 53% for text-based AI content.
A new paper from KTH argues coding agents have made traditional human code review obsolete. Here's what the argument actually says and where it breaks down.
Motion.so launches an AI agent for motion design that researches your brand, storyboards scenes, and outputs editable video from a text prompt.
PMB stores AI agent memory in a single SQLite file with hybrid retrieval via BM25 and vector search. No cloud, no API keys, runs fully offline with Ollama.
A developer's framework for when working AI-generated code still deserves rejection — and the specific patterns that trigger it during code review.
Airgap uses Linux namespaces to prevent AI coding agents and npm installs from accessing sensitive secrets like .env files and SSH keys.
Cloudflare now lets AI agents deploy Workers to temporary accounts with a single API call. Deployments live for 60 minutes, then get claimed or expire.
Elastic's engineering team details a persistent agent memory layer on Elasticsearch using three indices, hybrid retrieval with RRF, and per-user document-level security.
VLM's Orion 2 replaces sequential tool-calling with generated programs that execute end to end, cutting round-trips and latency for visual AI agents.
Open-source CLI tool aggregates token usage across Claude Code, Codex, Gemini CLI, and 12+ AI coding tools into one leaderboard. Free to use.
Bastion spins up isolated Linux VMs designed specifically for AI coding agents to run in the background without touching your local machine.
Stephen Bochinski's cost breakdown of AI coding tools for personal projects shows where the real expenses hide and how to keep them under control.
Anthropic is pulling Agent SDK and claude -p out of Pro and Max subscriptions on June 15, replacing the included usage with a fixed monthly API credit.