{"date":"2026-11-02","title":"What happened in AI — Nov 2, 2026","generated_at":"2026-11-02T22:16:00Z","intro":["Agent builders spent the day on containment: Uber's MCP gateway and four new sandbox, kernel, and signature tools all enforce permissions outside the prompt.","DeepSeek open-sourced a coding agent that starts work from team chats, and Nvidia published two DGX Spark inference efforts."],"highlights":["Uber's MCP gateway governs 800+ servers and 5,000+ agent tools.","Deedglass caught an agent reporting 1 updated row when Postgres changed 301.","DeepSeek open-sourced Super-Agent, a coding agent that watches team chats.","Google and Meta open-sourced TorchTPU for native PyTorch on TPUs.","SemiAnalysis launched GDPx and OpenAI launched LifeVal as agent benchmarks."],"article_count":35,"categories":[{"name":"Agent harness and context design","slug":"agent-harness-context-design","summary":"Builders are converging on one loop-plus-sandbox architecture, and the open problems have moved to context, memory, and cross-device state.","articles":[{"title":"New context engineering challenges in late 2026","summary":"LangChain argues the LLM-in-a-loop-with-tools-in-a-sandbox architecture has converged and lists the context-engineering problems that remain.","source":"langchain_blog","url":"https://www.langchain.com/blog/context-engineering-challenges-late-2026","published":"2026-11-02T16:51:00Z"},{"title":"Article: Systemic Approach to Memory, Knowledge, and Context in Agentic AI Architectures","summary":"InfoQ article lays out a systemic approach to memory, knowledge, and context in agentic architectures.","source":"infoq_ai_ml","url":"https://www.infoq.com/articles/agentic-ai-context-memory-knowledge/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=AI%2C+ML+%26+Data+Engineering","published":"2026-11-02T11:00:00Z"},{"title":"Show HN: Skep – live cross-device sessions for AI coding agents","summary":"Skep opens live join-code sessions so peers can share one task while each runs their own agent (Claude Code, Codex, Pi).","source":"hackernews_ai","url":"https://skepagent.com/","published":"2026-11-02T01:20:00Z"},{"title":"Show HN: AgentTM – A time-travel debugger for AI coding agents built in Go","summary":"AgentTM records AgentTM-managed Codex runs in SQLite with baseline and final workspace checkpoints, so you can replay what an agent changed.","source":"hackernews_ai","url":"https://github.com/Jyotishmoy12/AgentTM","published":"2026-11-02T08:54:58Z"},{"title":"Show HN: Rufous – Agent attuned to small models","summary":"Rufous is an agent harness built for 2B–12B models running locally, in an editor or a web UI.","source":"hackernews_ai","url":"https://rufous.chat","published":"2026-11-02T02:54:01Z"}]},{"name":"Agent permissions enforced outside the prompt","slug":"agent-permissions-outside-prompt","summary":"Four new tools and one gateway move agent access control into sandboxes, kernels, signatures, and verification instead of prompt instructions.","articles":[{"title":"Uber Builds MCP Gateway to Govern Access to 800+ Servers and 5,000+ AI Agent Tools","summary":"Uber built an MCP gateway to govern access to 800+ MCP servers and 5,000+ agent tools.","source":"infoq_ai_ml","url":"https://www.infoq.com/news/2026/11/uber-mcp-gateway-ai-agents/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=AI%2C+ML+%26+Data+Engineering","published":"2026-11-02T13:53:00Z"},{"title":"Show HN: Maat – AI agents that can diagnose Ubuntu but can't change it","summary":"Maat gives agents a kernel-confined MCP server for diagnosing Ubuntu; fixes need polkit approval and leave a tamper-evident audit trail.","source":"hackernews_ai","url":"https://github.com/bolesawny7/maat","published":"2026-11-02T13:53:51Z"},{"title":"Show HN: Deedglass – An agent said it updated 1 row; Postgres changed 301","summary":"Deedglass compares what an agent reported against what actually changed in Postgres (one reported row versus 301 changed) and seals each action as a signed record.","source":"hackernews_ai","url":"https://github.com/deedglass/deedglass","published":"2026-11-02T15:02:04Z"},{"title":"Lyngvi – A Hardened Sandbox for AI","summary":"Lyngvi runs coding agents in a hardened sandbox that never exposes credentials, git history, or a route to the internet.","source":"hackernews_ai","url":"https://lyngvi.skaperiet.no","published":"2026-11-02T21:31:49Z"},{"title":"Show HN: Arken – cryptographic identity for AI agents calling MCP servers/APIs","summary":"Arken proposes DID-signature identity so agents authenticate to MCP servers and APIs without passwords or sessions.","source":"hackernews_ai","url":"https://github.com/arkenforge/arken-core","published":"2026-11-02T04:29:23Z"},{"title":"Secure AI Agents with Hazmat Suits, Not Self-Defense Classes","summary":"Infisical argues agent security needs containment (the hazmat suit) rather than only trying to anticipate every attack.","source":"hackernews_ai","url":"https://infisical.com/blog/ai-agent-security-hazmat-suits","published":"2026-11-02T19:58:43Z"}]},{"name":"DeepSeek open-sources a proactive coding agent","slug":"deepseek-proactive-coding-agent","summary":"DeepSeek released a research prototype coding agent that picks up work by watching team chats instead of waiting for prompts.","articles":[{"title":"DeepSeek open-sources Super-Agent, a coding agent that watches team chats for work","summary":"DeepSeek open-sources Super-Agent, a coding agent that monitors team chats and starts work from them.","source":"search_cn_open_weight_labs","publisher_name":"Help Net Security","publisher_domain":"helpnetsecurity.com","url":"https://news.google.com/rss/articles/CBMidkFVX3lxTE5nREY5eGxfa2tXV0ZxeHZpekdIaS05UWdfYUxVTHNIR01TdnU0MEhxaHA2NE1iSHE2MnpHRUgtM3VvR2x2ZEpKQ1ljYVFEME9lQTEtOHNXSHdkT0Jjb0VPTUgtYnNVdEQ2RXBlMERLOTRTcF9ic2c?oc=5","published":"2026-11-02T07:09:55Z"},{"title":"DeepSeek team open-sources proactive coding agent research prototype","summary":"TechNode reports the release as a proactive coding-agent research prototype from the DeepSeek team.","source":"search_cn_open_weight_labs","publisher_name":"TechNode","publisher_domain":"technode.com","url":"https://news.google.com/rss/articles/CBMiowFBVV95cUxPZlhRZXhnYlJmX1RiOUdNamRqUXJUU2NycHNuZWtGeEVUejJxNVdEQkQzUU1HMkxHR1dXeDhBVDlPRmVyUHFoMU96VkdBYzlRUFRSWkFFTGVBWWFjNXRRaUlld2d2alpFdVd0VXFlR2tMeXNJUWdDc3c1cG5SR3BPRVd0OVRFLXBvMWhJZ0hzRG1qU2VQazBDQjNQNHl2cXRJZFZV?oc=5","published":"2026-11-02T04:32:41Z"}]},{"name":"Local and low-level inference","slug":"local-low-level-inference","summary":"Two Nvidia-hosted projects target DGX Spark's memory-bound decoding, and Google and Meta put PyTorch natively on TPUs.","articles":[{"title":"Optimizing 124B MoE inference on a single Nvidia DGX Spark","summary":"Nvidia profiles Ling-3.0-flash decoding on DGX Spark as memory-bound and targets the MoE computation and output-layer projection.","source":"hackernews_ai","url":"https://developer.nvidia.com/blog/3-ways-to-boost-moe-inference-on-nvidia-dgx-spark/","published":"2026-11-02T04:13:18Z"},{"title":"Flare, Nvidia's DGX Spark Inference Engine","summary":"Nvidia's Flare is a research-prototype local inference server for SM120/SM121 GPUs, aimed at long conversations and repeated tool calls.","source":"hackernews_ai","url":"https://github.com/NVIDIA/spark-local-inference","published":"2026-11-02T01:11:27Z"},{"title":"Google and Meta Open-source TorchTPU to Bring Native PyTorch Support to Google TPUs","summary":"Google and Meta open-source TorchTPU, bringing native PyTorch support to Google TPUs.","source":"infoq_ai_ml","url":"https://www.infoq.com/news/2026/11/torchtpu-native-pythorch-tpu/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=AI%2C+ML+%26+Data+Engineering","published":"2026-11-02T17:00:00Z"},{"title":"Lakebase Postgres is now generally available on Google Cloud","summary":"Databricks made Lakebase Postgres generally available on Google Cloud, with compute separated from storage for agentic workloads.","source":"databricks_blog","url":"https://www.databricks.com/blog/lakebase-postgres-now-generally-available-google-cloud","published":"2026-11-02T14:26:08Z"}]},{"name":"Benchmarks and compute economics","slug":"benchmarks-compute-economics","summary":"New benchmarks target economically valuable agent work, while the compute-glut debate leans on DeepSeek V4.1 Flash pricing.","articles":[{"title":"GDPx: Measuring Frontier Model Capabilities Across the Most Valuable Domains","summary":"SemiAnalysis introduces GDPx, a benchmark of the highest-value domains for frontier agentic capability.","source":"semianalysis","url":"https://newsletter.semianalysis.com/p/gdpx-measuring-frontier-model-capabilities","published":"2026-11-02T01:51:42Z"},{"title":"LifeVal","summary":"OpenAI's LifeVal benchmarks AI on everyday tasks such as finding childcare and handling government paperwork.","source":"openai_blog","url":"https://openai.com/index/lifeval","published":"2026-11-02T09:00:00Z"},{"title":"Jevons paradox is lazy analysis: Evaluating the risk of AI overbuilding","summary":"A widely shared estimate argues next year's online compute would need about $2.5T of revenue even at DeepSeek V4.1 Flash rates, and the author is revising it.","source":"hackernews_ai","url":"https://twitter.com/bookwormengr/status/2116705246856741114","published":"2026-11-02T09:16:24Z"},{"title":"Introducing the Grid Protectors Program","summary":"OpenAI's Grid Protectors Program works with utilities so future compute campuses protect grid reliability and consumer electricity costs.","source":"openai_blog","url":"https://openai.com/index/grid-protectors-program","published":"2026-11-02T15:00:00Z"}]}],"video":{"youtube_id":"bIspSsxwJrU","title":"LLM Digest Daily Explainer: Nov 2, 2026","duration_s":112.6}}