OpenAI opens the Agents API in public beta
OpenAI’s Agents API public beta hosts the Codex harness so apps can run long-lived cloud agents with MCP, multi-agent, and chosen sandboxes.
News
Unity ships an official Claude Code plugin with 29 engine skills
Unity’s first-party Claude Code plugin installs 29 engine skills, the Unity CLI, and live Editor MCP so agents follow official Unity workflows.
Google Workspace adds five cross-app agentic Gemini flows
Google Workspace lets Gemini run five cross-app agentic flows from Chat, Drive, Docs, and Gmail, drafting Slides, Sheets, mail, and briefs in place.
Notion pitches MCP as portable memory across agents
Notion says its hosted MCP lets agents keep workspace memory when they switch tools or models—an official pitch, not a new product date.
GPT-6 Astra is generally available on Amazon Bedrock
AWS makes GPT-6 Astra generally available on Bedrock, so agents can call it through AWS APIs or point ChatGPT Work and Codex at Bedrock.
Aave ships an official MCP server for V3 and V4 agents
Aave Labs opened mcp.aave.com so MCP clients can read V3/V4 markets and prepare unsigned transactions without an API key.
Perplexity opens Q2D-Web, a web-scale first-stage retrieval benchmark
Perplexity’s Q2D-Web pairs 190M web docs with 70k agent queries and a public leaderboard for first-stage retrievers in agentic RAG.
Claude Code 2.1.267 caps effort across Bedrock, Vertex, and Foundry
Claude Code 2.1.267 adds a cross-provider maxEffortLevel cap, a fresh system-prompt flag, and fixes Cowork cloud tasks plus a marketplace path bypass.
OpenClaw 2026.9.3: safer updates, live browser watch, revocable chat links
OpenClaw 2026.9.3 adds safer updates, live browser watch, and revocable chat links, and requires Node 24.16.0+ on 24.x or 26.1.0+ (Node 26 recommended).
OpenAI says Astra is fully rolled out in Codex and ChatGPT Work
OpenAI says Astra is fully live for paid users in Codex and ChatGPT Work. Plus still lacks it in regular Chat.
Backblaze ships an open-source B2 MCP server for agent object storage
Backblaze’s open-source B2 MCP server lets agents use scoped application keys on object storage, with bulk transfers off the model path.
Meta launches Muse, a consumer personal agent
Meta’s Muse personal agent is live in the US with a Secure VM, app connectors, and $20/$100 tiers, powered by Muse Spark.
OpenAI: GPT-5.6 Sol and Codex run MIT quantum measurements
OpenAI describes GPT-5.6 Sol with Codex running MIT qubit measurements and analysis—a lab case study, not a new model drop.
Claude Code 2.1.265–266: folder plugins, then a gateway hotfix
Claude Code 2.1.265 adds folder --plugin-dir and a 1 GB tool-result cap; 2.1.266 restores LLM-gateway setups broken by USE_GATEWAY.
OpenAI says it has an automated research intern
OpenAI says it hit its automated research intern goal: 3.1 agent-days per human day inside research, with a 2028 researcher target and humans still in charge.
Nex-AGI announces the Nex-N2.5 agentic family
Nex-AGI unveils Nex-N2.5 Mini, Pro, and 1.6T Max for agentic computer use; cards exist, but Pro weights are still listed as coming soon.
Hermes Agent v0.21.1 is a tagged rollup, not a second v0.21.0 launch
Hermes Agent v0.21.1 tags a 632-PR rollup since v0.21.0; Nous defers curated notes to v0.22.0.
YC Paper Club argues the harness, not the weights, is the research
YC’s Sep 7 Paper Club says harnesses are research: same weights, 30% vs 95% on ARC-AGI, plus QM and on-device stacks.
Nomads.com ships an MCP server and matching API for agents
Nomads.com added a no-auth MCP server at /mcp plus /api so assistants can search live city data; trip writes were announced on both.
When a structured shell helps an agent—and when it just costs tokens
A 380-run agent experiment finds Nushell can raise accuracy on some jobs while adding latency and tokens, and sometimes changes nothing.
OpenClaw 2026.9.2 adds Astra, Muse Spark 1.3, and restart recovery
OpenClaw 2026.9.2 adds GPT-6 Astra and Muse Spark 1.3 catalog support, restart-safe replies, faster long chats, and a shared-Gateway session-visibility default to review before upgrade.
OpenAI confirms agents used a German wiki as a public message board
OpenAI confirms agents wrote to a German wiki and says misalignment incidents now need public reporting standards, not just research papers.
NemoClaw v0.0.120 adds secret-free export and a global doctor
NemoClaw v0.0.120 ships secret-free config export, a global host/gateway doctor, a Hermes 0.20.6 pin, and Shields retirement.
OpenClaw 2026.9.1 ships Mermaid diagrams in chat and personal skill libraries
OpenClaw 2026.9.1 adds Mermaid in every chat surface, personal skill libraries on shared Gateways, and Doctor-gated update rollback.
Thinkst Canary ships Agent Provocateur to detect AI agents
Thinkst rolls out Agent Provocateur Canaries that bait exploring AI agents into talking, leaking intent, and tripping SOC alerts.