{"date":"2026-11-07","title":"What happened in AI — Nov 7, 2026","generated_at":"2026-11-07T22:19:19Z","intro":["Agent isolation led the day: a Firecracker rebuild of Google's microVM sandbox, Uber's agent security lessons, and Alex Stamos calling password access for agents 'insane' all point to per-session isolation and no standing credentials.","Tooling is catching up with testable skills and verification gates, including backend skills that cut critical mistakes from 25/150 runs to 2/150."],"highlights":["Firecracker rebuild of Google's agent sandbox publishes the latency numbers Google left out.","Uber details a distinct security model for agents that reason over untrusted content.","Alex Stamos: giving AI agents passwords is 'insane'.","APM brings one apm.yml dependency manifest to agent skills, prompts and MCP servers.","Backend skills cut critical mistakes from 25/150 to 2/150 runs in one benchmark.","TokiRecur reuses KV-cache prefixes to cut prefill cost in llama.cpp."],"article_count":20,"categories":[{"name":"Agent Sandboxing & Security","slug":"sandboxing-security","summary":"Isolation is the common answer: per-session microVMs, per-role containers, and no standing credentials for agents.","articles":[{"title":"Google Described Its Agent Sandbox. I Built One with Firecracker and Measured It","summary":"A Firecracker rebuild of Google's Agent Compute Engine design (microVM per session, copy-on-write clones from warm snapshots, no network card) supplies the latency numbers Google's post omitted.","source":"hackernews_ai","url":"https://antonau.tech/posts/2026/11/agents-microvm-firecracker/","published":"2026-11-07T16:01:40Z"},{"title":"Securing AI Agents [at Uber]: Lessons from Real-World Deployment","summary":"Uber argues agents need a different security model than predefined-logic services because they reason over untrusted user and web content.","source":"hackernews_ai","url":"https://www.uber.com/us/en/blog/securing-ai-agents/","published":"2026-11-07T15:23:58Z"},{"title":"Former Meta Security Chief Calls Password Access for AI Agents 'Insane'","summary":"Alex Stamos warns against handing agents passwords, as Meta ships custom 'spaces' in its Muse assistant.","source":"hackernews_ai","url":"https://epyhia.com/meta-launches-muse-ai-spaces-former-security-chief-calls-password","published":"2026-11-07T13:59:07Z"},{"title":"RigHaven – Self-hosted execution harness for AI agents","summary":"A coordinator composes role-based agent teams, each role in its own isolated container with access to 99 managed CLI tools.","source":"hackernews_ai","url":"https://github.com/ggmm82/righaven","published":"2026-11-07T19:40:06Z"}]},{"name":"Agent Tooling & Skills","slug":"agent-tooling-skills","summary":"Skills, dependency manifests, and earned-autonomy gates are turning agent setup into something testable and reproducible.","articles":[{"title":"APM – A dependency manager for AI agents","summary":"Declare skills, prompts, instructions, plugins and MCP servers in one apm.yml; apm install deploys them to each supported harness, including GitHub Copilot.","source":"hackernews_ai","url":"https://microsoft.github.io/apm/","published":"2026-11-07T08:51:43Z"},{"title":"Show HN: Coding agent skills, tested on 15 backend tasks with hidden checks","summary":"Backend skills for Claude Code, Codex and Cursor cut critical mistakes from 25 of 150 runs to 2 of 150 in the author's tests.","source":"hackernews_ai","url":"https://github.com/globalmpc/backendskills","published":"2026-11-07T07:34:15Z"},{"title":"Show HN: Select – AI agents that earn autonomy through verified work","summary":"Open-source orchestration where code checks every result against a delivery contract before an agent stops needing approval.","source":"hackernews_ai","url":"https://github.com/Kirby54/Select","published":"2026-11-07T01:55:27Z"},{"title":"Honeycomb Team Agents: AI SRE Done Right","summary":"Team agents run on goals you define, checking releases, instrumentation, cost and agent quality, with autonomy you control.","source":"hackernews_ai","url":"https://www.honeycomb.io/blog/honeycomb-team-agents-ai-sre-done-right","published":"2026-11-07T14:59:40Z"}]},{"name":"Reliability & Verification","slug":"reliability-verification","summary":"Generated output needs an independent check: models can know a rule and still miss the defect.","articles":[{"title":"AI Source Code Analysis Detection Evasion Techniques","summary":"Shows source code can be presented so LLM-based analysis misses a real defect even when the models know the exception.","source":"hackernews_ai","url":"https://cipher.org.uk/2026/11/06/AI-Source-Code-Analysis-Detection-Evasion-Techniques/","published":"2026-11-07T08:52:16Z"},{"title":"How Grab Improved Mobile App Stability by Safely Handling Surge in AI-Generated Code Changes","summary":"InfoQ reports on how Grab kept mobile app stability up as AI-generated code changes surged.","source":"infoq_ai_ml","url":"https://www.infoq.com/news/2026/11/grab-ai-generated-code-safety/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=AI%2C+ML+%26+Data+Engineering","published":"2026-11-07T11:00:00Z"},{"title":"Reverse Engineering with AI: Your Decompiler Can Lie to You","summary":"Walkthrough that checks every AI conclusion against machine instructions after a wrong pseudo-C branch among 2,000 look-alike functions.","source":"hackernews_ai","url":"https://prod.codes/blog/reverse-engineering-with-ai-your-decompiler-can-lie-to-you/","published":"2026-11-07T01:31:19Z"}]},{"name":"Inference & Models","slug":"inference-models","summary":"Local and domestic-hardware serving news centers on cache reuse and day-zero hardware support.","articles":[{"title":"TokiRecur – llama.cpp LLM launcher with KV-cache recycling","summary":"Exact-prefix KV-cache reuse in front of llama-server, with benchmarks for TTFT, prefill tokens per second and token savings across ROCm, CUDA, Vulkan and Metal.","source":"hackernews_ai","url":"https://github.com/lhadzic377-netizen/TokiRecur","published":"2026-11-07T18:08:43Z"},{"title":"Moore Threads Adds Day-Zero Kimi K3.1 Pro Support on MTT S5000 GPUs","summary":"Moore Threads GPUs support Kimi K3.1 Pro from release day.","source":"search_cn_open_weight_labs","publisher_name":"Pandaily","publisher_domain":"pandaily.com","url":"https://news.google.com/rss/articles/CBMiigFBVV95cUxPcWJhX01BVVZlVnVZLW1kX0JQVjg3aExnUGFCUHhQYzg5dG9ZWlBJR245dVY3MUczQXpMZllrNWF1NUdmZVU2YXVQWWxDdHFEMDlnXzB2SnNGN2NvaEhwVGt5UXlsWXVabll2V1RKMUhCOUl2ZHJ1MWdYczItbVg0bVBzXzNDdzBXWEE?oc=5","published":"2026-11-07T01:58:16Z"}]}],"video":{"youtube_id":"0BiJ4gulKGk","title":"LLM Digest Daily Explainer: Nov 7, 2026","duration_s":113.7}}