Skip to content
#Open source / repos
TodayNovember 13Friday29 items
  1. Hacker News · AI58

    Mandat launches authorization, delegation and audit layer for AI agent tool calls, MCP first

    Mandat is an open-source (Apache-2.0) tool that adds authorization, delegation and audit to AI agent actions, starting with MCP. Each tool call is checked against a policy on its arguments, such as allowing refunds up to €50, requiring human approval for €50–500 and denying anything above €500, while agents receive short-lived single-task tokens carrying the human to agent to sub-agent chain and every call leaves a tamper-evident audit record exportable to a SIEM. It also ships a scanner for authentication and supply-chain mistakes in MCP servers and client configs, and installs via npm or pip with TypeScript and Python libraries for the official MCP SDK.

  2. Hacker News · AI65

    Brainfish AI open-sources Webcrawl, a zero-key web crawling API and MCP server for agents

    Brainfish AI released Webcrawl, an open-source web crawling API and MCP server that lets agents such as Claude Code, Cursor and Codex search, scrape, map, crawl and monitor sites without an API key. One command, npx webcrawl setup, registers the @webcrawl/mcp server and installs the webcrawl-anything Agent Skill, while npx skills add brainfish-ai/webcrawl-anything installs all eight skills, including ticket, job, deal and real-estate watchers. The same TypeScript engine runs locally and in the hosted webcrawl.app cloud, and the project is MIT licensed.

  3. MarkTechPost58

    Google releases TPU Sync, an open-source KV cache transfer library for Cloud TPUs

    Google Cloud has released TPU Sync, an Apache-2.0 C++ library that moves KV caches and model weights across TPU HBM, pinned host DRAM, shared storage and the data center network. In internal July 2026 tests on TPU v7x, Google reports over 90% lower KV transfer overhead, with Qwen3 235B median KV transfer latency dropping from several seconds to about 230 ms and average TTFT falling 60% for 235B and 90% for 480B at low concurrency. The library integrates with PyTorch via TorchTPU and with JAX, serving as the KV connector in vLLM-TorchTPU and the P/D transfer backend in SGLang-TorchTPU, though the README says it is not yet recommended for general use.

  4. Hacker News · AI36

    Agent-brain: A Notes-Agnostic Second-Brain Operating Model for AI Coding Agents

    Agent-brain is a personal operating model for AI coding agents such as Claude Code, OpenCode, Codex and Antigravity CLI, built on a second-brain structure of journal, WIP, memory and tasks that works over any folder of notes, with Obsidian optional. Runtime config and memory like CLAUDE.md and AGENTS.md are kept in a git-tracked brain and symlinked into each runtime, so setup travels across machines. Install runs via curl and dry-runs by default, with a session lifecycle driven by the brain skill.

  5. Hacker News · AI58

    Witness launches a voice-anchored evidence recorder for AI coding agents

    Witness is a voice-anchored evidence recorder that links each spoken comment to the clicks, requests, console errors, DOM mutations and rrweb replay of that moment, then hands the bundle to an AI coding agent. It runs as a CLI (npm i -g witness-recorder, Node 18+) that opens a managed Edge/Chrome browser, and it exposes the session through a stdio MCP server with tools such as list_segments, events_around and get_dom_state for agents like Claude Code, Cursor and Devin. Recordings stay local in .witness/sessions, password fields are masked and body capture can be disabled with WIT_NO_BODIES=1.

  6. inclusionAI Models34

    inclusionAI Releases Duo-1.0-distill-quant FP8 Weights for Full-Duplex Interaction

    inclusionAI released Duo-1.0-distill-quant, FP8-quantized four-step distilled Realization weights for the Duo 1.0 full-duplex multimodal world model. The weights overlay onto the inclusionAI/Duo-1.0-causal-base bundle, which supplies the shared Intent, codecs, speech models and Realization backbone, and require a runtime supporting the packaged ModelOpt FP8 format. The weights were initialized from MiniMax H3 and then trained by the Duo team.

  7. inclusionAI Models40

    inclusionAI Releases Duo-1.0-distill Four-Step Distilled Weights for Duo 1.0

    inclusionAI has released Duo-1.0-distill, the four-step distilled Realization weights for Duo 1.0, a full-duplex multimodal world model for real-time audio-visual interaction. The weights ship as a PyTorch Distributed Checkpoint overlay on the inclusionAI/Duo-1.0-causal-base bundle, using an Euler sampler with 4 steps and flow shift 5.0. The weights were initialized from MiniMax H3 and then trained by the Duo team.

  8. Hacker News · AI58

    Study finds AI-style stale version pins in 13.7% of new npm packages in 2026 H1

    Kobi Mozgori sampled 250,000 npm packages and found that new packages whose pinned dependency versions are mostly round x.0.0 numbers already superseded rose from a 1–2.4% baseline in 2019 to early 2024 to 13.7% in 2026 H1, falling to 7.6% in 2026 H2 through early November. The median pinned dependency was about 155 days old at birth in 2021–2023 but 451 days in 2026 Q2, and the share of packages born with essentially all pins on the newest release dropped from about 28% to about 10%. The author frames the pattern as an inference rather than proof, since registry metadata cannot show who or what wrote a manifest, and notes the decline since mid-2026 is unexplained.

  9. Hacker News · AI26

    Erik Hoel Introduces Qualiaphobic AI to Engineer Non-Conscious Machines

    Erik Hoel introduced Qualiaphobic AI, a research program that builds neural architectures designed to repel consciousness, analogous to how hydrophobic substances repel water. The proposal explores three approaches: distilling, transforming or ablating AI against popular consciousness theories, shifting architectures toward "anchor points" like lookup tables, and fragmenting potential phenomenology. Hoel plans to construct and open-source these techniques, arguing they could resolve AI welfare and make humans the only moral patients.

  10. Hacker News · AI58

    ChatPanel Releases Chap: Generating Meeting Notes from Live Captions, Supporting Any LLM or Agent

    ChatPanel launched Chap, a browser, desktop and terminal meeting assistant that turns live captions from Zoom, Meet, Teams and Webex into running notes, decisions and action items without a bot joining the call. It runs on the user's machine with no ChatPanel server in the path, anonymizes emails, phones, cards and keys before requests leave the browser, and lets users switch mid-conversation between an in-browser WebGPU model, Ollama or LM Studio, 20+ cloud APIs, and agents such as Claude Code, Codex, GitHub Copilot and Antigravity. Free covers the first 10 notes and 10 meetings; Pro is a limited-offer $49 one-time purchase or $42 per year.

  11. Hacker News · AI58

    Uber Open-Sources Michelangelo ML Platform, Says It Has Been Validated Across 400+ Projects and 20,000 Monthly Training Jobs

    Uber has open-sourced Michelangelo, its ML platform proven at Uber scale with 400+ active ML projects, 20K+ model training jobs per month, 5K+ models in production and 45M+ real-time predictions per second at peak. The release covers feature management, distributed model training, deployment and monitoring, with feature management, evaluation and monitoring marked coming soon. A Python API built on Uniflow lets users define distributed training tasks and end-to-end workflows, such as a Ray-based XGBoost training pipeline, in a few lines of code.

  12. Hacker News · AI58

    Zipzi launches open-source Live2D Cubism Editor alternative with AI agent co-editing

    Zipzi released an alpha open-source 2D modeling and animation editor positioned as an alternative to Live2D Cubism Editor, built in Rust with GPUI for macOS and Windows. A Python SDK (CPython 3.14) lets a coding agent inspect meshes, render wireframe overlays, and apply edits in the same document as the user, with shared undo history. The project is in early development with no stable release, and the SDK is MIT-licensed while editor components use GPL-3.0-only.

  13. Hacker News · AI48

    Loose Ends: A liveness check for unused code AI agents leave behind

    Loose Ends is a new CLI tool that checks whether code added or touched in a change is still reachable from a real entry point, flagging orphaned additions, newly orphaned code, and replaced implementations left in place. It builds a liveness graph across TypeScript, JavaScript, Python, and Java, and reports only what the change did, so existing dead code never blocks a pull request. Installable via npm, it requires Node.js 20 or newer and git.

  14. inclusionAI Models38

    SingVL-4B GGUF Released: GGUF Build of Sing VL S1 4B Multimodal Decision Model on Qwen3.5

    Sing VL S1, a 4B-parameter multimodal decision model built on Qwen3.5, now has a GGUF release for llama.cpp CPU inference. It ships in BF16 (8.42 GB), Q8_0 (4.48 GB) and Q4_K_M (2.71 GB) decoders plus a shared F16 vision projector, with a default 4096-token context. GGUF validation covers 9 examples and 11 questions; full public scores for the GGUF variants are still pending.

  15. Hacker News · AI38

    LessonFolk: Open-Source AI Tutor That Teaches AI via MCP

    LessonFolk is a free, open-source AI tutor that teaches AI topics through the AI chat app you already use, connected via MCP. It offers 15 courses, from an hour-long AI Foundations track to Context Engineering for Long-Running Agents, with a short level check and progress tracking. Courses and code are open source, and users can download or delete their data anytime.

  16. Hacker News · AI58

    How much of GitHub's most-starred AI tooling is written by coding agents?

    An analysis of the full git history of 52 of GitHub's most-starred AI-tooling repositories, covering 324,088 commits, finds that in the median repo 16.9% of commits in its last 180 days of activity carry an AI signature, and 21 of 52 sit at 25% or more. Stars and AI-signed share show no detectable link (Spearman ρ = -0.05, p = 0.72), and 85.3% of the 36,892 signed commits come from Claude, whose tool adds a trailer by default. The author cautions that heavy AI use is concentrated in projects started in 2025 or later, so the sample cannot separate AI use from recency, and every AI percentage is a lower bound.

  17. Hacker News · AI71

    InferSSD runs 800B-class LLMs on consumer hardware by keeping expert weights on NVMe

    InferSSD, an open-source inference runtime built on llama.cpp, runs 763B-class DeepSeek-V4.1-Flash-1017 and 816B-class Muse-Spark-Open on consumer machines by keeping routed expert weights on NVMe storage while attention, shared layers and the KV cache stay in RAM or VRAM. On an RTX 5090 with 32 GB RAM and 32 GB VRAM it reports 164.5 tokens/s prefill and 7.27 tokens/s decode for DeepSeek at a 4K prompt tier, versus 8.06 and 0.867 tokens/s for llama.cpp on the same system. The article also estimates that heavy daily use could mean roughly 85 TB of host reads per day and lays out a conditional SSD wear model, noting the figures are assumptions to test rather than a promised drive lifespan.

  18. Hacker News · AI76

    AgentRun Open-Source Release: Adding Self-Improvement and Durable Execution to Existing Agent Harnesses

    Parcha Labs has open-sourced AgentRun, an agent platform that layers self-improvement, durable execution, and a console on top of Claude Code, Codex, OpenCode, and pi, under the Apache-2.0 license.

    Why it matters: AgentRun layers self-improvement, durable execution, and a console on top of existing harnesses such as Claude Code and Codex, and provides a concrete case where code review costs dropped from $6,451 per month to 18 cents per run.

  19. Hacker News · AI38

    OwnOwl v0.2: On-Device LLM Chat App for iPhone Test Build

    OwnOwl released a v0.2 test build of its iPhone app that runs a large language model entirely on-device, with no internet used for chatting. The update speeds up first replies by warming up on load and caching each conversation, adds longer default answers with a Continue button, and lets the app remember user-confirmed facts. It requires an iPhone with 6 GB of RAM or more and iOS 17 or newer, with model downloads of about 6 GB (GigaChat), 9 GB (Ling), or 14 GB (Gemma).

  20. Hacker News · AI58

    CanyonOS 0.2 launches as open-source tool to find, observe and scale agents

    CanyonOS 0.2 is the first public release of an open-source, model-agnostic tool for finding, observing and scaling agents that slow down workflows. It runs on Docker or Kubernetes, works with any provider offering an OpenAI-compatible API, exports OpenTelemetry-compatible traces and metrics to tools such as Datadog, Grafana, Prometheus, Langfuse and Arize Phoenix, and adds cost and queue time per agent. The project is AGPL-3.0 licensed, requires agents to be rewritten into plain Python with one file per agent, and its FAQ states it has not yet run at large production scale.

  21. Hacker News · AI58

    Stay-in-sight launches open-source outbound email tool with 64 MCP tools for AI agents

    Stay-in-sight released an MIT-licensed outbound email system that runs on Cloudflare Workers and D1, sending cold outreach, newsletters and drip series from the user's own Gmail and SES mailboxes. It ships 64 MCP tools plus a REST API so agents like Claude Code and Cursor can import contacts, schedule sends and answer replies, and the project claims deployment in about 30 minutes with hosting free at low volume or $5/mo on Workers Paid.

  22. Hacker News · AI58

    Relate launches an open-source embedded graph database for AI agents

    Relate is an open-source embedded graph database for AI agents that runs inside the application with no server to deploy and no signup, storing data in memory, a directory or a single .rl file. It speaks openCypher and is tested against the openCypher TCK, and ships with an agent skill plus an MCP endpoint so agents such as Claude Code, Codex and Cursor can write Cypher queries against connected facts. Installation runs through a curl script that verifies release signatures and SHA256 checksums, and libraries are offered for Rust, Python, JavaScript, WebAssembly, Go, Java, Swift and C.

  23. Hacker News · AI60

    Cerulion 2.0 released: an open-source ROS 2-compatible robot runtime with multi-process replay

    Cerulion Inc. released Cerulion 2.0, an open-source, ROS 2-compatible robot runtime that records runs and replays failures against changed code, with one process per node by default. It ships Linux x86_64/ARM64 and macOS binaries, an rmw_cerulion layer tested on Foxy through Lyrical, and MCAP recordings with byte-exact replay verification that returns a nonzero exit code on divergence. On one x86-64 Linux desktop at 1 MiB payload, the project reports a 4.08 µs median round trip versus 10,952 µs for ROS 2 Jazzy with rmw_fastrtps_cpp defaults.

  24. Hacker News · AI88

    Edge0 open-sources Edge0-145B and Edge0-120B on-device models with inference framework

    Edge0 open-sourced Edge0-145B and Edge0-120B along with its unified inference framework and deployment tools, claiming Edge0-145B is the first open-source 145B-parameter on-device model. The company reports 29GB peak memory, 1,100+ tokens/s prefill, 80+ tokens/s decode and roughly 120W typical power on a compatible laptop, with 128K–256K context. It cites internal scores of 64.6 on DeepSWE 1.1, 86.52 on Terminal-Bench 2.1, 90.4 on GPQA-Diamond and 80.6 on IFBench, and says weights are on Hugging Face with day-one support for macOS, Windows and Linux.

    Why it matters: Edge0 open-sources a 145B on-device model that reports 29GB peak memory and 80+ tokens/s decode, offering a concrete data point on how far local inference can scale.

  25. MarkTechPost52

    Liquid AI Releases LiquidCodec, a 74M-Parameter Streaming Audio Codec

    Liquid AI released LiquidCodec, a roughly 74M-parameter streaming speech tokenizer that powers its LFM3-Audio-2B and LFM3-Omni-2B models, with weights published on Hugging Face. The codec compresses 24 kHz audio into 12 token streams at 12.5 Hz frames and runs on a single CPU thread, where Liquid AI reports about 361x real-time encoding and 87x decoding. Liquid AI says it beat X-Codec 2, NanoCodec and Mimi on ViSQOL and mel distance across all 8 test sets, though NanoCodec still led PESQ (WB) on 7 of 8 sets.

  26. Hugging Face Blog69

    IBM releases LLMON preview models built on Granite 4.1 3B and 8B

    IBM released two experimental preview models, granite-4.1-3b-LLMON-preview and granite-4.1-8b-LLMON-preview, derived from Granite 4.1 base checkpoints and trained to consume and generate LLMON, an LLM-native markup language that marks instructions, data and requirements with reserved special-token delimiters. On the OPI benchmark the models avoid 47% more indirect prompt injection attacks than the best frontier model tested and 95% more than the best open-source model, while performing on par with Granite 4.1 Instruct on standard benchmarks. The release includes a GitHub toolkit with lossless JSON-to-LLMON converters, and the authors note the preview is for exploration, not production.

    Why it matters: IBM trained a preview model on Granite 4.1 that can read and write LLMON tags, using explicit instruction-data separation to reduce the indirect prompt injection success rate on the OPI benchmark to near zero while maintaining general capabilities.

  27. MarkTechPost71

    AWS Releases AWS Emulator, a Local Emulator of 200+ AWS Services Written by AI Agents

    AWS has released AWS Emulator, a free local emulator of AWS APIs that runs on a laptop, dev server or in CI without an AWS account, and all of its code was written and reviewed by autonomous agents. It is built from 438 AWS API models covering 200+ services and passes 24,243 of 42,857 scored fidelity tests (56.6%) in release v1.0.322, with AWS KMS at 93.9% and AWS CloudFormation at 15.8%. AWS lists clear limits: no IAM policy evaluation, one region and one account, in-memory short-lived resources, and telemetry on by default.

  28. AWS Machine Learning Blog60

    AWS open-sources UAEF, a framework-agnostic evaluation system for AI agents

    AWS has open-sourced UAEF (Universal Agent Evaluation Framework), a framework-agnostic system that uses adapters to normalize agent traces from LangGraph, LangChain, Strands Agents, Amazon Bedrock AgentCore and Langfuse into one canonical AgentTrace, then scores them with 36 built-in metrics across 7 dimensions. Results are stored per experiment in Amazon DynamoDB and Amazon S3, so teams can compare agent versions against a baseline, flag regressions with a default 5 percent threshold, and generate statistical reports plus LLM-written insights. The post says customers across 12 organizations in telecom, healthcare and financial services used UAEF before release, and the Apache 2.0 code, Python library, CDK deployment and per-framework notebooks are in the awslabs/uaef GitHub repository.

    Why it matters: The AWS team has open-sourced UAEF, which uses adapters to unify agent trajectories from different frameworks into a single format, then scores them across 7 dimensions using 36 metrics, providing a reusable evaluation layer for comparing agent versions across frameworks.

  29. AWS Machine Learning Blog58

    AWS walks through building cross-organization multi-agent systems with A2A

    AWS published a reference architecture in which a LangChain-based enterprise orchestrator on Amazon ECS calls a Strands Agents-based compliance SaaS agent on Amazon Bedrock AgentCore over the A2A protocol, with the two sides in separate AWS accounts and no shared secrets. Each side authenticates with tokens from its own Amazon Cognito user pool, and the SaaS agent calls back to an enterprise context agent for account metadata, so the agents act as peers rather than client and server. The system deploys from three Jupyter notebooks that provision five CloudFormation stacks and run an end-to-end test plus seven trust boundary checks.

November 12Thursday
  1. Databricks Blog58

    How Block uses Buzz and Databricks Unity Gateway to run collaborative AI agents

    Block built Buzz, an open source collaboration workspace where employees and AI agents work together in group discussions, focused conversations and one-on-one exchanges, and uses Databricks Unity Gateway as the model access layer. Employees can create agents, pick the model or coding harness behind each one, and select Databricks as their inference provider to reach frontier models from OpenAI, Anthropic and Google plus select open-weight models through a single integration. Block engineers are also experimenting with multi-agent setups where a more capable model coordinates agents running smaller, more efficient models, and have built an internal approach to benchmarking these agent teams.

  2. AWS Machine Learning Blog36

    Build autonomous AI agents with Amazon Bedrock and Spring AI

    AWS shows Java teams how to build autonomous AI agents using Spring AI and Amazon Bedrock's AgentCore, with a spring-ai-agentcore integration. Two reference systems are detailed: a single-agent travel assistant with tool calling and memory, and a multi-agent intelligent document processing platform with five specialist agents, a rules engine, and alerting. The stack requires JDK 21, Spring Boot 4.1 with Spring AI 2.0, and deploys via AgentCore runtime on ARM64 containers.

  3. Hugging Face Blog36

    Hugging Face Releases Fine-Tuning Guide and Scripts for TRELLIS.2 3D Generation

    Hugging Face published fine-tuning scripts and containers for TRELLIS.2, a 3D generation model, runnable locally or on Hugging Face Jobs. The guide uses LoRA adapters via the PEFT library on only the sparse-structure flow DiT and shape flow DiT, leaving the material stage frozen. On a 188-object footwear subset of the GSO dataset, the pretrained model often rendered incorrect soles, which the LoRA fine-tune targets.

  4. AWS Machine Learning Blog34

    Semantic diffing of formal policies with anti-unification

    An open source engine computes semantic diffs of formal policies using anti-unification over an e-graph, showing shared structure once and keeping both versions of each differing part. It targets autoformalization outputs in SMT-LIB, Cedar, Dogwood, OpenShell policies and CEL, where an LLM writes the artifact but plays no part in computing the diff.

  5. AWS Machine Learning Blog36

    What landed for AI builders on AWS in October 2026: Claude Haiku 5.5, GPT-6.1 Terra, AgentCore CLI 1.0

    AWS's October 2026 roundup adds Claude Haiku 5.5, which Anthropic says costs around 75 percent less than Claude Haiku 4.5 for most tasks, plus OpenAI GPT-6.1 Terra on Amazon Bedrock with up to 1 million input tokens and prices at least 50 percent lower than GPT-6.1 Sol and GPT-5.6 Terra. AgentCore CLI reached version 1.0, and AgentCore Gateway added model fallback and Step Functions workflows as agent tools.

November 11Wednesday
  1. MarkTechPost60

    MBZUAI's IFM Releases PAN 2, an Open 29B World Model for Physical AI

    MBZUAI's Institute of Foundation Models released PAN 2, a 29B-parameter open world model that takes a scene plus a text action and simulates the outcome as video, with weights, training code and inference code under Apache 2.0. It pairs an LLM planner that writes structured JSON world states with a 29B MMDiT diffusion transformer conditioned by a frozen Qwen3.5-9B encoder, and IFM reports 34.0% full-plan completion on 100 multi-step tasks, ahead of MiniMax-H3 at 32.50%, while trailing MiniMax-H3 on single actions and human preference (Elo 1595 vs 1620). IFM's ablation shows removing the LLM planner drops full-plan completion to 0.0% and step completion from 75.1% to 8.7%.

  2. Hugging Face Blog42

    Tether AI Research Releases BotPsy: A 0.8B On-Device Agentic Model Built on Qwen3.5-0.8B

    Tether AI Research released BotPsy, a 0.8B-parameter agentic model built on Qwen3.5-0.8B for on-device personal-assistant tool use. It roughly doubles its base model's agentic scores, with BFCL v4 rising from 23.4 to 44.0 and τ-bench from 23.2 to 51.5, and beats every model up to 1.2B parameters tested across six tool-use benchmarks. It also cuts unsafe responses to 21% from 36% and uses about 3× fewer reasoning tokens on GPQA Diamond.

  3. inclusionAI Models34

    inclusionAI Releases LemonDecider 4B v3 Decision Model Based on Qwen3.5-4B

    inclusionAI released LemonDecider 4B v3, a decision model fine-tuned from Qwen3.5-4B that returns choice, yes/no or ordinal probabilities via a single joint forward pass instead of generating answer tokens. It was trained on about 1.20M decision/QA rows and scored 61.67 on the public Decision Index 0.4 suite, with median request latency of 49.9 ms.

  4. inclusionAI Models40

    inclusionAI Releases ZoomGain-9B and ZoomGain-4B Models with ZoomGainBench and RL Training Data

    inclusionAI released the ZoomGain-9B and ZoomGain-4B vision-language model checkpoints, alongside the ZoomGainBench evaluation benchmark and ZoomGain RL training data. ZoomGain mines global-to-zoom perception gaps to build fine-grained visual supervision, and ZoomGain-9B reached superior performance on multiple fine-grained and general perception benchmarks. ZoomGainBench contains 1,000 questions over 941 images, while the RL dataset holds 9,000 training records and 7,713 source images.

  5. inclusionAI Models38

    inclusionAI Releases ZoomGain-4B and ZoomGain-9B with ZoomGainBench and RL Training Data

    inclusionAI released the arXiv paper, ZoomGainBench evaluation benchmark, RL training dataset, and ZoomGain-4B/ZoomGain-9B model checkpoints. The best of 28 models scored 61.08 on ZoomGainBench, while ZoomGain-9B reached superior performance on multiple fine-grained and general perception benchmarks. ZoomGain mines global-to-zoom perception gaps to build fine-grained visual supervision for vision-language models.