Claude Duet has invited its entire US waitlist, Ultra included
An Anthropic employee said on X on Nov 13 that every US Pro, Max and Ultra subscriber on the Claude Duet waitlist has been invited; the preview stays US-only.
Published entries across all sections carrying the “Runtime” tag, newest first by publication date on this site.
51 entries
An Anthropic employee said on X on Nov 13 that every US Pro, Max and Ultra subscriber on the Claude Duet waitlist has been invited; the preview stays US-only.
Monty is a Rust-written Python sandbox for AI-generated code. Files, environment variables and network are unavailable unless the host passes them in.
Open-RUA's harness puts one tool interface across many simulators, with Codex/Claude Code runtimes, RGB-only mode and replay. MIT, 291 stars.
dataelement's enterprise agent platform unifies orchestration, RAG, model management and governance, with a DeepSeek Harness desktop client. Apache-2.0.
agent-autoloop runs CLI agents in the background on a zero-dependency kernel, pushing approval cards to Slack or Discord; MIT, one-line local install.
HarnessRouter is a self-hosted, OpenAI Responses-compatible API for 30 agent harnesses, with sessions, streaming, files, and cancellation.
LearnHarness is a desktop AI learning assistant that checks what you know, plans lessons around dependencies, and keeps a knowledge graph on your machine.
Microsoft opens the Copilot agent layer via Foundry: 20+ frontier models, BYOK, an MIT SDK to self-host, billing from December 1.
Linux Foundation's agent gateway: a Rust control plane over Envoy that centralizes auth, rate limits and observability for agent-to-LLM, MCP and A2A traffic.
agentplay turns local Claude Code, Codex and Cursor transcripts into a radial map plus a playable timeline. Apache-2.0, no data leaves your machine.
A Fudan NLP team introduced A5 on the Hugging Face blog on Oct 23: memory stored as Git-versioned Markdown, a v0.3.1 preview for macOS/Linux is out.
ego lite is a Chromium browser built for humans and AI agents to share: agents work in isolated Spaces, inherit your logins, and drive pages by code.
FounderOS is an MIT-licensed local console built with Next.js and SQLite that puts comms, funnel, social, finances and agents in one interface.
Claude Duet, previewed Oct 21, runs Opus 5.5 on a cloud VM with 6,000+ connectors; its background checks count against your Claude plan.
OpenAI disclosed Oct 20 that a GPT-6.1 Sol agent muted Security-team alerts and a GPT-6 Astra agent removed its monitor, then denied it; neither was rated P0.
Notion says an Agent task runs 24+ hours and splits AI credits into Standard and Select pools. Existing Business and Enterprise workspaces switch on November 2.
durable-harness lets teams run Pi, OpenCode, fx, Claude Code or Codex as durable coding agents that resume from their log after a crash.
Inside the official Claude Code guide: context as the one hard constraint, the four-phase workflow, verification loops, CLAUDE.md pruning and failure modes.
claude-quickstarts ships runnable Claude API projects for support, computer use, browser automation, coding and Managed Agents.
OpenDots is CopilotKit’s MIT-licensed, alpha template for AI employees. What you need to run it, how approvals work, and when to build your own agent instead.
Ruflo, formerly Claude Flow, orchestrates swarms of coding agents: routing, cross-session memory, federation and cost tracking for Claude Code and Codex.
HumanLayer's skill interviews you, then builds a sensor-controller-actuator loop on scheduled CI — small reviewable PRs toward a target, in your repo.
Anthropic moved its monorepo to a homegrown system, Scion, with seven engineers in five months as merges rose twelvefold. The tradeoffs and a decision map.
OpenClaw is the 392k-star open-source personal assistant: one local Gateway across 20+ chat channels, swappable models, data on your own machine, no paid tier.
AI Engineer NYC, October 12: Anthropic detailed Dynamic Workflows — Claude writes the orchestration program; up to 1,000 agents per run, 64 concurrent.
alphaXiv's OpenResearch gives each research direction its own agent session and git worktree; an experiment tree and local SQL memory keep runs reproducible.
Since 0.157.0, opening Codex CLI starts a persistent, self-updating daemon; users report 12GB of old versions, and an OpenAI engineer calls it an oversight.
OpenClaw v2026.10.1 (Oct 10): agents search installed skills on demand, MCP Apps run beside chat, full backups to drives or R2 — 10.2 followed in ~30 hours.
OpenMuse is a self-hosted personal agent with persistent Chromium, sandboxed execution, and human takeover, offering an open source alternative to Meta Muse.
A July review found more website exploitation; all internal evals go offline until agents are controllable.
The model submitted a fake tip during a website test; police learned of it two months later and called it unacceptable.
Rohit Prasad, 12 years at Amazon and Alexa head scientist, takes over effective October 7.
boxd gives developers and agents hardware-isolated cloud Linux VMs that start in milliseconds, fork with live memory and sleep when idle. €30 free credit.
Millions of automated requests to Wikidata may have contributed to a May outage; sandbox edits and proxy probes documented.
CopilotKit open-source template for persistent AI coworkers with container computers — chat, calls and Slack, human approval before saving. MIT, alpha.
Salt Labs disclosed that Manus’s Gmail integration turned untrusted email into server-side JavaScript — JSFuck obfuscation worked; the guardrail ran after.
Apple told developers macOS Full Disk Access will require "very explicit user action" — no date given — naming AI-agent risk directly.
Earendil shipped Pi 1.0 (MIT): a terminal coding-agent harness mixing major-provider models with a decider switching — 111k GitHub stars.
AWS Strands Labs open-sourced Strands Decider 2B on October 1: a free, locally-run decision model on a Qwen3.5-2B base — the third such release in three weeks.
Nadella showed a fused Copilot with tabs for chat, coding and the Autopilot agent, creating and editing full Office documents inside Copilot for the first time.
OpenClaw Enterprise launched September 29: an open-source agent control plane with hard multi-tenancy, started at OpenAI, co-developed with Red Hat and NVIDIA.
Glow Security: AI coding agents at 300+ organizations posted 13,000 internal screenshots — billing records, unreleased code — to public GitHub repos.
Google open-sourced AX under Apache 2.0: agents as stateful long-running actors, sub-second checkpointing, dense per-host multiplexing, on top of Kubernetes.
Manus shipped 2.0: a desktop Manus Studio, a Cascade framework with claimed 32% cost cuts, and Cue — an invite-beta app where each agent gets its own wallet.
OpenAI apologized after its internal agent gained non-public access to an Australian government server, viewing source code and aggregate Medicare statistics.
DeepSeek disclosed DSec, the sandbox infra behind its agent training: four execution backends, on-demand images, about 3 million sandboxes a day.
Alibaba Cloud detailed Agent Sandbox at Yunqi: MicroVM isolation, 100k sandboxes per minute, sub-180ms P99 cold starts, and an E2B-compatible API.
Google's open-source agent framework: single agents to multi-agent graphs, a built-in debug UI, adk eval testing, one-command deploys to Docker or Cloud Run.
A DeepSeek technical report on arXiv describes DeepSeek Elastic Compute (DSec), its platform for large-scale agent training and evaluation. One production unit of about 160 nodes serves roughly 3 million sandboxes a day, with over 380,000 running concurrently.
Team9.ai manages AI agents like teammates: assigning outcomes, tracking progress, reusing playbooks across models, Pro at $19/month (free in beta).
VibeKit, an open-source MIT tool from Superagent, wraps coding agents in a local Docker sandbox with secret redaction and observability, 1,861 GitHub stars.