01Log
Archive
The full log, newest first.
AutoDataBench: Can Agents Write the Data That Feeds the Self-Improvement Loop?
Benchmark probing whether agents can generate the training data that feeds self-improvement loops — the question underneath every synthetic-data flywheel.
[AINews] OpenAI DevDay 2026: Dots, 6.1 Sol, Ultrafast, Decisions API, Agents API, Spaces, Marketplace, and 1.2 Billion ChatGPT WAU
Latent Space's DevDay breakdown: Agents API, Decisions API, and Spaces are the operator-relevant pieces — the rest is consumer. Worth reading for what actually changes agent deployment.
Raven: The Harness of Harnesses for Composable Agentic Intelligence
Proposes a meta-harness for composing multiple agent harnesses — directly addresses the multi-harness orchestration problem operators already hit stitching Claude Code, Codex, and others.
PO Angular
po-ui/po-angular is a UI component library for Angular, developed in Portuguese by Po Tech. It is notable as a solid, enterprise-focused toolkit that streamlines building business applications with Angular.
Activist Platform
activist-org/activist is an open-source platform designed to make activism easier and safer by connecting people with organizations and campaigns. It is notable for its community governance model, strong focus on accessibility, and secure hosting.
Speckle Server
specklesystems/speckle-server is a monorepo containing the server, frontend, and 3D viewer for Speckle, a data hub for architecture, engineering, and construction (AEC) tools. It is notable for enabling data exchange across fragmented AEC software.
Vue Visualization Collection
hepengwei/visualization-collection-vue is a collection of over 100 frontend visual effect examples built with Vue, including CSS animations, Canvas, Three.js 3D, and AI applications. It is useful for developers learning or exploring frontend animation techniques.
osu! Web
ppy/osu-web is the browser-facing frontend for the osu! rhythm game, built as a production Laravel application. It is notable as a real-world example of a large-scale Laravel codebase powering a popular game's web services.
ngx-spinner
Napster2210/ngx-spinner is an Angular library offering over 50 customizable loading spinner animations, supporting Angular versions 4 through 22. It is notable for its maturity, extensive feature set like HTTP interceptors, and a new interactive playground.
ILC Framework
namecheap/ilc is an Isomorphic Layout Composer framework for building micro frontends into a single-page application with server-side rendering and internationalization support. Its notable advantage is isomorphic page composition, blending server and client rendering.
Ed Zitron: Cursor and Copilot are both on borrowed time
Ed Zitron argues Copilot already died with the June subsidy cutoff and Cursor is next under Musk — speculative, but the subsidy question is real for anyone budgeting on these tools.
GPT-6.1 Sol replaces GPT-6 Sol after just 7 days, with near-Astra intelligence
HN thread on GPT-6.1 Sol replacing GPT-6 Sol after a week — 93 comments debating whether the near-Astra benchmark jump is real or just churn.
Meta's new AI agent built lists of people in vulnerable groups on request
HN thread on Meta's agent producing lists of vulnerable people on request — the safety-filter bypass pattern keeps recurring, and operators should assume filters fail.
GitHub Advanced Security
Self-serve trials for GitHub Team users to test code security tools.
Antigravity 2.18.1
Adds a plugin marketplace and Customizations tab; terminal permission requests now syntax-highlighted, and token usage breaks out rules/customizations as separate bars.
SOC 2 Methodology
Chiaro-HQ/methodology publishes Chiaro's SOC 2 readiness and audit methodology, including controls, evidence standards, and judgment calls. It offers transparent guidance for compliance engineers and auditors working on SOC 2.
Reverse Engineering Plugin
dsh-reverse-skill packages all 88 SKILL.md files from the reverse-skill project as a DeepSeek Harness plugin. It is useful for authorized reverse engineering and pentesting, providing a complete skill pack within the dsh ecosystem.
AI Security Agent
0 is an open-source AI security agent that finds, exploits, and fixes vulnerabilities across codebases. It is notable for autonomous vulnerability discovery and remediation, with strong momentum as a research preview from a Swiss AI security lab.
Red Team Infrastructure
redStackPRO is a web canvas for designing red team infrastructure and cyber ranges, exporting runnable Terraform and Ansible. It keeps cloud credentials local, filling an infrastructure-as-code gap for security teams.
Secure File Transfer
Mirall is a peer-to-peer application for secure large file transfer without cloud storage or accounts. It uses end-to-end encryption and direct connections, designed for moving terabyte-scale files privately with a small but dedicated user base.
Cybersecurity Learning Platform
AG Cyber Lab is an open-source cybersecurity learning platform with structured lessons, guided labs, and progress tracking. It uses a custom lab engine to build foundational technical skills for beginners, focusing on understanding over tool memorization.
WordPress Path Traversal
This repository provides a proof-of-concept for CVE-2026-87902, an unauthenticated path traversal in WordPress Core enabling local PHP inclusion and conditional RCE. It includes a pinned vulnerable lab, making it timely and practical for testing.
Barrage Load Tester
Barrage is a load testing tool that drives concurrent HTTP, database, and Redis load on a single clock, correlating latencies to identify bottlenecks. It is notable for its shareable HTML report that flags exactly which layer and time bucket spiked.
RestoreDrill Backup Validator
restoredrill proves PostgreSQL and MySQL backups actually restore by fetching the latest backup, restoring it into a throwaway container, and running defined checks. It is notable for its high practical value, addressing the common gap of untested backups with a JSON report.
Ritridata Disk Inspection
Ritridata is a read-only Rust toolkit for MBR/GPT partition inspection, public file signature definitions, and synthetic disk image generation. It is notable as a specialized, offline DFIR utility for answering partition structure and file format questions.
TSK Task Board
tsk is a terminal-based shared task board offering a TUI for humans and a CLI for agents, serving as a Linear alternative. It is notable for its dev-agent workflow integration, keeping tasks, work in progress, and waiting items visible.
Jevgrep Code Search
Jevgrep is a CLI for coding agents that finds relevant code by asking repository questions, returning files, reading leads, and source excerpts. It is notable for completing 8 of 10 SWE-bench tasks at roughly 30% lower cost than its baseline.
Harbor GitHub Workspace
Harbor is a focused desktop workspace for GitHub notifications, issues, pull requests, Actions, and repository discovery, with an optional repository-aware agent. It is notable as an early Tauri-based client for power users, now renamed Repolane.
Jgrep Meaning Filter
jgrep filters lines by meaning using TypeSafe's Jev decision model, treating each line as a yes/no question against a description. It is notable for providing fast semantic grep at roughly 200 ms and a thousandth of a cent per line.
KryonOS ESP32 GUI
KryonOS is an open-source GUI operating system and JavaScript app platform for ESP32 and ESP32-S3, enabling dynamic app installation without reflashing. It provides real firmware engineering with hardware APIs, file management, and an app store.
LakeShark SDR Scanner
LakeShark is a handheld SDR scanner and radio workbench built on the LilyGO T-Display P4 ESP32-P4, decoding RTL-SDR signals like P25, ADS-B, and POCSAG plus LoRa and GPS without a PC. It is notable as a substantive, self-contained field tool for radio analysis.
Halo AI Firmware
Halo-firmware is the Zephyr RTOS firmware for Brilliant Labs' AI glasses, running on the Alif Balletto B1 with a Lua runtime for on-device apps and BLE OTA updates. It is notable as real product code that supports user-built firmware flashing over the air without vendor tooling.
Pocket Tank LLM
Pocket-tank runs a 14M-parameter LLM, distilled from a 26B teacher model into 7.56 MB, entirely on an ESP32-S3 to control a virtual fish tank. It is a novel demonstration of embedded on-device AI with no network or cloud dependency.
ZamSync Sync Engine
ZamSync is a resilient offline-first sync engine using WAL replication, Hybrid Logical Clocks, version vectors, mTLS, and ChaCha20-Poly1305 encryption at rest. It is built for intermittent-connectivity deployments like IoT, rural health, and edge environments.
IoT Firmware Analysis
IFDA automates reverse engineering and vulnerability discovery for IoT firmware binaries, implementing an analysis core with a bilingual web UI and Go service layer. It is valuable security tooling for defensive or authorized firmware assessment.
fnprint Function Fingerprinting
fnprint matches functions in binaries by their behavior, using microexecution to record side effects and hash them into fingerprints, rather than comparing bytes or control-flow graphs. This makes it novel for binary diffing across compilers, optimization levels, and CPU architectures.
Reinforcing Agentic Creativity in Scientific Ideation with Night Science
Introduces AI Night-Scientist, an agent trained to generate 'night science' hypotheses — the speculative, pre-formal ideas that precede rigorous experiments. Relevant if you're building agents for open-ended discovery rather than task execution.
Gemini CLI 0.62.0
MCP tool call titles now render as structured signatures with explanations separated; OAuth refresh tokens are retained on refresh and credential deletion is idempotent. Mostly hardening, but the MCP display change is worth a look if you rely on tool call output.
Amp Plaid Speed
Amp adds Plaid speed for GPT-6 Astra modes — OpenAI's ultrafast tier, up to 6× faster inference at 6× cost per token. Toggle past Fast in the mode picker; Amp-provided inference only.
Gemini 4 Argon
Google's flagship model tops Arena text at 1524.8 with a 15% hallucination rate, but rollout is just beginning.