UnYOLO is a credential broker and policy engine for AI agents accessing GitHub. Agents get short-lived proxy tokens instead of real credentials.
Hermes Agent routes inference through Vercel AI Gateway (200+ models, no token markup) and executes commands in Vercel Sandbox's isolated microVMs. Setup is a picker flow.
Figma's custom skills let designers encode reusable instructions for the AI agent, turning repeated prompts into shared team knowledge across Figma Design and Make.
Vercel ships @ai-sdk/workflow 1.0.0 with stable telemetry, Node 22 minimum, and a default block on system messages in agent prompts.
Polygraph gives AI coding agents visibility across multiple repos and persistent session memory, tackling the context fragmentation problem in multi-repo codebases.
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.
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.
Firecrawl is now available on the Vercel Marketplace, giving teams structured web data for AI agents without managing crawling infrastructure.
GitHub launched a standalone Copilot desktop app at Build 2026 with parallel agent sessions, git worktree isolation, and a new Agent Merge workflow.
Microsoft released a free, open-source LangChain.js course with 70+ TypeScript examples covering agents, MCP, and agentic RAG for JavaScript developers.
Cursor 2.5 adds a plugin system, async subagents that run tasks in parallel, and a sandbox.json networkPolicy that enables per-domain network access for sandboxed commands.
Microsoft published an architecture for securing MCP tool calls with identity, authorization, validation, rate limits, and audit logging. Here's what it means.
Cursor 2.0 introduces a proprietary coding model and a redesigned agent sidebar with plan-based workflows. Here's what shipped and how to use it.
Cursor's new long-running agents can plan and execute complex coding tasks autonomously over extended periods, now available as a research preview.
Cursor shipped self-hosted cloud agents that keep code inside your infrastructure. Here's what changed, who it's for, and how to evaluate it.
Cursor announced an SDK for building programmatic coding agents. Here's what the changelog reveals, what's unconfirmed, and what to do next.