AI Tools Daily

MiniMax releases M3, an open-weight model with a 1M-token context window — and we teach how sparse attention makes that affordable. Plus Claude Code 2.1.166 adds fallback models and strips user authority from agent-to-agent messages (the confused deputy problem, explained), and Cursor 3.7's Context Usage Report finally shows where your tokens go. Hosts: Alex & Jules. New episodes daily.

MiniMax launches M3, an "open-weight" frontier coding model with a 1M-token context — but the weights aren't out yet and the benchmarks are self-reported. Plus: GitHub Copilot CLI's new adversarial /rubber-duck command and scheduled prompts, Claude Code's security-hardening release wave, and Cursor's new tiered Teams pricing. Hosts: Alex & Jules. New episodes daily.

Windsurf got an over-the-air update and opened as Devin Desktop — Cognition unified its cloud and local agents under one IDE with ACP support for Claude, Codex, and more. Plus: what Claude Opus 4.8's Dynamic Workflows actually unlocks for multi-agent builders, Claude Code's new auto-loading plugin system, and GitHub Copilot's $100 Max plan. Hosts: Alex & Jules. New episodes daily.

Microsoft Build dropped a wave of GitHub Copilot CLI updates yesterday: a Rubber Duck adversarial review agent, /every and /after prompt scheduling, local voice input, Gemini model support, and a GA SDK for building Copilot extensions. Plus a breaking change in Claude Code's changelog that silently kills workflow hooks — and a community knowledge graph tool claiming a 120x token reduction over sta...

GitHub Copilot's billing flips to AI Credits today — no more fallback model when you run out — and we break down exactly what changes and what to check. Plus: Claude Code now auto-loads plugins from .claude/skills directories (no marketplace needed), auto mode lands on Bedrock/Vertex/Foundry, Windsurf's SWE-1.6 is free for three months with meaningful agent UX improvements, and Hugging Fac...

Claude Code dropped two major updates overnight: Opus 4.8 with high-effort defaults and a new dynamic workflows system that orchestrates tens to hundreds of background agents in parallel. We also cover Zyphra's open-source MoE diffusion model hitting 7.7x inference speedup, and LiteLLM shipping a Kubernetes-based self-hosted agent platform for production teams. Hosts: Alex & Jules. New epi...

Anthropic ships self-hosted sandboxes and MCP tunnels for Claude Managed Agents — letting agents run tool execution on your own infrastructure and reach private networks without exposing endpoints. Claude Code v2.1.152 adds `/code-review --fix` to auto-apply review findings to your working tree. Cursor launches Composer 2.5, a frontier-level in-house model matching Opus 4.7 benchmarks at one-tenth...

Google officially retired Gemini CLI at I/O and gave free, Pro, and Ultra users until June 18 to move to Antigravity CLI — but the switch from open-source to closed-source has developers frustrated. We cover that, plus a Claude Code `/code-review` command that now posts inline GitHub PR comments, a critical macOS crash fix in Claude Code, and GLM-5.1, a new open-weight coding model with an MIT lic...

Anthropic acquired Stainless — the SDK generator used by OpenAI, Google, and Cloudflare — and is winding it down for everyone but Claude. We break down what that means for developers on both sides. Plus: Google's new Managed Agents API turns one API call into a full sandboxed Linux agent loop, xAI enters the coding CLI race with Grok Build, and the inevitable convergence of every AI coding too...

Anthropic's biggest 2026 update lands: Claude Opus 4.7 hits 87.6% SWE-Bench, plus Dreaming (self-improving agent memory), Outcomes (rubric-based auto-retry grader), and multiagent orchestration in the console. Google I/O's Antigravity 2.0 and WebMCP origin trial redefine browser-based agent workflows. Plus: Claude Code /usage breakdown, Cursor Composer 2.5's frontier-beating price, Win...

Cursor shipped Composer 2.5 — its first in-house coding model — matching Claude Opus 4.7 benchmarks at a fraction of the price, and it signals a bigger independence play from the Anthropic ecosystem. We also cover the Google I/O developer drops that went beyond the headlines, Claude Code's new `/code-review` command, and a handy CLI for taming MCP config sprawl across multiple AI coding tools....

OpenAI open-sourced Symphony, a SPEC for orchestrating fleets of coding agents through your issue tracker — a simple but powerful architecture shift for teams hitting the multi-session ceiling. Cursor shipped Composer 2.5, built on Moonshot's open-weight Kimi K2.5 model and matching Opus 4.7 on SWE-Bench, plus a new Jira integration that brings Cursor into where enterprise teams already work....

Google I/O 2026 delivered: Firebase AI Logic hits general availability, Gemma 4 drops as an open-weight local model, and Firebase Studio enters the agentic coding race with MCP integration. Meanwhile, Anthropic's acquisition of Stainless officially closed — the SDK generation startup that OpenAI and Google both depend on. Plus: OpenAI's new Codex Spark text-only model just entered research...

Anthropic launches four new managed agent capabilities from their Code w/ Claude SF event — including Dreaming (agents that improve between runs), Outcomes (rubric-based self-revision), and multiagent orchestration — alongside Claude Code rate limits that just doubled. Windsurf makes GPT-5.1 its free default this week, and Hugging Face's open-source ml-intern agent (9.3k stars) autonomously tr...

Anthropic is splitting Claude subscriptions: starting June 15, programmatic usage via claude -p, GitHub Actions, and third-party agents moves to a separate credit pool billed at API list prices — and the math is steep. Also: Claude Code v2.1.143 ships strict plugin dependency enforcement and context cost estimates, and Google I/O's agentic coding keynote (Firebase Studio) lands Tuesday. Hosts:...

Claude Code 2.1.0 dropped with 1,096 commits — hooks now live inside agents and skills, forked context isolates skill execution, and /teleport sends your terminal session to the browser. Also: Hugging Face's ml-intern open-sources the ML post-training loop, and GPT-5.5 lands as the new default in Codex CLI. Hosts: Alex & Jules. New episodes daily.

Anthropic ships 10 open finance agent templates as Claude Code plugins — and the reference architecture is worth studying for any builder. Google's Chrome DevTools team drops an official MCP server letting coding agents inspect and control a live browser. Plus: why letting your agent click around a website costs 45x more than using the API directly, OpenAI's new default ChatGPT model (GPT-...

Cursor just shipped version 3 with a brand-new agentic interface and cloud VMs — but the old IDE isn't going anywhere. Plus: Anthropic's Code with Claude event revealed "Dreaming," a research-preview feature that lets Claude inspect its own sessions overnight and write its own playbooks. We also cover Claude Code Routines for CI auto-fix, the new `claude project purge` command, a...

Anthropic's Code w/ Claude event delivered real developer features: Claude Code Routines enable async PR-filing agents, "Dreaming" lets managed agents learn from their own sessions overnight, and rate limits just doubled thanks to the new Colossus data center deal. Also: eight xAI Grok models are being retired on May 15 with barely two weeks notice — action required now. Plus Qwen3.6...

Claude Code ships another fast-follow release with MCP visibility fixes that will save developers real debugging time. Plus: n8n-mcp hits 16k GitHub stars as the bridge between AI coding tools and workflow automation, Rekall brings a native C++ MCP server inside Unreal Engine with 480+ tools, and GPT-5.5 launches as the model parity era makes tool ecosystem fit the deciding factor. Hosts: Alex & J...

Anthropic's auto mode for Claude Code is now rolling out to Max users — a background AI classifier that handles permission decisions so you don't have to. Plus: Cursor 3.1 canvases let agents generate live React dashboards in your editor, Cursor's new enterprise model controls let admins lock down providers org-wide, Claude Code v2.1.126 adds a project purge command and gateway model picker, and R...

A critical command injection flaw affects over 200,000 MCP servers — and Anthropic says it's working as intended. Plus: Windsurf 2.0 ships Devin as a built-in cloud agent, Google's Agent Development Kit hits stable v1.0 with cross-framework A2A protocol support, and Hugging Face open-sources an ML engineer agent that automates post-training end-to-end. Hosts: Alex & Jules. New episodes...

Claude Code 2.1.126 lands with a new `claude project purge` command for cleaning up project state, plus Opus 4.7 is now the default model on Max and Team with a new xhigh effort level and /ultrareview out of preview. Google's Chrome DevTools MCP gets live browser session attachment via --autoConnect, Cursor 3's /best-of-n runs parallel implementations in isolated worktrees, and OpenCode hi...

A critical, by-design remote code execution vulnerability in Anthropic's MCP SDK puts 200,000 servers at risk — and Anthropic says it won't patch it. Plus: OpenAI's GPT-5.5 is now in Codex and on AWS Bedrock, Microsoft Agent 365 goes generally available for enterprise agent governance, and Qwen drops two open-source releases including an LLM interpretability toolkit. Hosts: Alex &...

Cursor 3.2 ships /multitask — a mode where the IDE decomposes your task and hands it to async subagents automatically. Claude Code v2.1.126 adds gateway model routing and project purge. Google's Chrome DevTools team releases an official MCP server for live browser debugging inside coding agents. Plus: Hugging Face's ml-intern automates the entire LLM post-training loop autonomously. Hosts:...