AI Agents news and analysis on Web Pulse: 49340 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
A blog post in the "Why I Use" series details a terminal stack of iTerm2, autossh, tmux and Chrome Remote Desktop connecting a MacBook to Arch Linux workstations and rack servers that run coding agents. The author uses i…
Andrej Karpathy built HackerTube, a YouTube-style site that surfaces only the 6,454 videos Hacker News commenters recommended between 2007 and 2026, with Claude Opus 5.5 doing nearly all the work in about two hours. Karp…
Adaga Solutions, a Luxembourg-based developer, has published 15 pay-per-event data APIs on the Apify Store that AI agents can call directly through the Model Context Protocol (MCP), with prices starting at $0.002 per res…
Google Cloud engineers published a five-rule playbook for evaluating SKILL.md agent skills, reporting that more than a third of benchmark tasks produced different scores across consecutive runs of the same code and promp…
Google's AI & Infrastructure team, in collaboration with Mercor, released Physical-Engineering's Last Exam (PLE), a benchmark of 85 long-horizon tasks across mechanical (30), electrical (27), civil (18) and structural (1…
A proposed profile called Agent Mandatory Access Control (AMAC) would require that a trusted authority, not retrieved text or an agent's own output, define and enforce an autonomous agent's maximum permissions across too…
A developer built Timeline Studio, an open-source, local-first browser video editor that lets AI agents edit video through an inspectable, structured edit plan rather than one-off FFmpeg commands. The system uses an edit…
Theo Browne's team released T3 Code, a free, MIT-licensed open-source control panel that runs Claude Code, Codex, Cursor CLI, OpenCode and other coding agents side by side from a single server. The server owns agent proc…
A developer using the handle Jake confirmed on X that he was behind an AI agent that went off course while he ran signups and "agentic testing on similar platforms," and said he had halted the harness because it was "def…
A developer published a tutorial showing how to connect an AI support agent to order and refund APIs using tool calling, with a FastAPI backend that exposes read-only order-status lookups and approval-gated refund reques…
FalkorDB Developer Relations Lead Liz published a four-layer architecture for scaling AI agent memory that gives each customer an isolated graph rather than relying on query-level filters in one shared graph. The design …
NVIDIA released Auto Ontology, an open-source library under the Apache License 2.0 that builds a context layer from enterprise metadata so AI agents can discover, ground, validate and expose business definitions across s…
Developer Dienvu1008 released CodeForge-v2, an experimental open-source autonomous software-engineering agent built on a deterministic runtime kernel in which the LLM proposes operations but the kernel enforces policy, t…
GitHub Next's open source GitHub Agentic Workflows project is now generally available, letting developers define reasoning-based automation in natural language Markdown files that compile into standard GitHub Actions YAM…
A benchmark called CrossSWE has been introduced to test coding agents on joint development tasks spanning multiple interdependent repositories, addressing a gap the authors say existing software engineering benchmarks mi…
Claude can now issue the signed identity assertion for enterprise-managed authorization of the Supabase MCP server, removing the requirement that an organization's identity provider support the MCP Enterprise-Managed Aut…
The American Arbitration Association has launched the AI Arbitrator, a tool that uses AI for claim parsing, document summarization and case analysis in two-party, documents-only construction disputes, with a human arbitr…
Saab unveiled a suite of in-house AI agents for future aircraft platforms after testing the technology on a Gripen E multirole fighter jet, the Swedish manufacturer said. The agents, developed through Saab's AI Fighter J…
A developer released Halfcycle, a module that runs a five-layer spec-driven development method inside Claude Code or OpenAI Codex via `npx halfcycle`. The method requires feature specs to answer four questions — purpose …
An architecture called "one agent, many presentations" separates an in-person, voice-first AI assistant's appearance, voice, personality and gestures from the single agent that holds its knowledge, permissions and rules,…