01Log
Archive
The full log, newest first.
Iris 3B
Small enough to run on your own hardware for asset generation when you would rather not call a hosted image API.
AgentGarten: Code Worlds for Evolving Agents
Agents learn through interaction, and what they can learn is bounded by the environments they learn in — AgentGarten's answer is code worlds built for evolving agents.
Step 5 Preview
A flagship agentic model aimed at long-horizon tool use and multi-step workflows.
Claude Haiku 5.5 in GitHub Copilot
Worth reading
Understanding and Enhancing Backdoor Persistency in LLM Agent Post-Training
Asks whether benign post-training can remove inherited backdoors in LLM agents, and examines how persistent those backdoors remain — security reading for anyone fine-tuning or sourcing agent models.
CheckerBench: Can Long-Horizon Agents Synthesize Static-Analysis Checkers?
Introduces an executable benchmark for evaluating whether long-horizon agents can synthesize static-analysis checkers — a concrete yardstick for coding-agent evals on extended, multi-step work.
Beam
A tiny, teachable WebGPU library that exposes the real WebGPU API through five nouns and two verbs with first-class TypeScript types.
Viewi
A front-end framework that compiles PHP components into reactive browser apps, worth exploring if you want to build interactive web UIs without leaving PHP.
hairline
A dependency-free ESM library that renders isometric line figures responding to pointer movement, usable with React or any DOM-based app.
Quickdraw
An MIT-licensed infinite-canvas whiteboard SDK that embeds a complete drawing surface into React, React Native.
Coi
A type-safe, component-based language that compiles to WebAssembly for building reactive web apps.
Strands Decider 2B: a small, open-source, decision model
270 points and 76 comments on HN for Strands' small open-source 2B decision model; one commenter who says they regularly use specialized micromodels warns, 'you won't be happy with such a model.'
Slop Alert: Super Mario Sunshine PC port has Claude code
A Bluesky 'Slop Alert' post (23 likes, 11 reposts) claims the Super Mario Sunshine PC port has Claude code in it.
OpenAI's Decisions API is in public beta — HN discussion
376 points and 219 comments on HN for the Decisions API public beta; the thread already includes a curl example calling the v1/decisions endpoint.
Cursor: service degradation — Cloud Agents and Grok Bot
Service degradation affecting Cloud Agents and Grok Bot — Grok Bot was restored first while the Cloud Agents investigation continued; resolved after 159 minutes.
GitHub: Incident with Git Operations, Pull Requests and Actions
Incident spanned Git Operations, Pull Requests, and Actions, resolved after 71 minutes — re-check any Actions runs queued in that window.
Anthropic launches expanded Cyber Verification Program
The expanded CVP makes advanced cyber capabilities and reduced blocking classifiers available to qualifying security professionals, now organized into three access tiers that security teams can apply for.
Cursor: Remote control for local agents
You can now see and reply to the local agents running on your computer from Cursor — remote control for agents that stay on your machine.
OMP 18.8.0
Breaking: the env API-key helpers (getEnvApiKey, getEnvApiKeyName, listProvidersWithEnvKey) are no longer exported from @oh-my-pi/pi-ai/stream — import from @oh-my-pi/pi-ai or @oh-my-pi/pi-ai/env-api-key — and TranscriptionRequest.audio now accepts Uint8Array | Blob, so consumers must handle Blob. Plus CPU/memory efficiency gains across streamed responses (Cursor, Devin, Codex, OpenAI Responses, GitLab Duo Workflow) and faster tool-output pruning/telemetry in long-running sessions.
VulnHunter
VulnHunter is an Apache-2.0 AI security scanner maintained as a harness-portable fork of Capital One's tool and used to prove vulnerabilities with sandboxed
action-dist-verify
action-dist-verify is an MIT-licensed TypeScript utility that rebuilds a pinned action's dist from source and diffs the committed artifact
Witness Vault
Witness Vault is an MIT-licensed Python tool that packages secrets into a self-decrypting offline HTML vault with AES-256-GCM WebCrypto
Hot Cell
HotCell is an MIT-licensed Ruby sidecar for Rails apps to run untrusted code on untrusted inputs in an unprivileged container
Persistent Browser-In-The-Middle
P-BitM is a GPL-3.0 Python platform for controlled browser-in-the-middle security assessments with an admin dashboard, isolated browsers, session viewing
Pared Apple Intelligence Cleaner
A tool that selectively removes unwanted Apple Intelligence models on macOS without disabling SIP, via a native app, CLI, nix-darwin, and Home Manager. It also underpins RemoveMacAI, which credits it for the hard reverse-engineering work.
Plannotator Terminal Markdown Annotations
A terminal Markdown annotator for selecting text and leaving comment, looks-good, or delete marks, then handing the numbered review to a coding agent. A single static Rust binary aimed at the current wave of agent-assisted code review workflows.
Tsuzuri Terminal Notebook
A block-based notebook for the terminal that stores notes as plain Markdown and offers Vim-style editing, live preview, boards, calendars, and 96 themes. A single-binary install for macOS, Linux, and Windows makes it an unusually full-featured TUI note-taking option.
Dopbase DevOps Secrets Manager
An open-source, self-hosted secrets manager packaged as a single file containing the server, Admin UI, REST API, and CLI, storing data in SQLite. A practical, still-maintained option for small DevOps teams wanting to keep secrets on their own infrastructure.
RemoveMacAI macOS Debloater
A native macOS app and CLI to disable Apple Intelligence, delete its models, and silence analytics, ads, pop-ups, and background updaters, with every change previewable and reversible. It builds on pared's reverse-engineering and drew press coverage, though its star growth looks inflated.
Terrahour Terminal World Clock
A terminal world clock that renders a braille-dot day/night world map plus a 24-hour per-city timeline for planning meetings across zones. It includes DST handling and time scrubbing, and is a polished, fresh single-tool TUI.
PuppetLoom
A Windows desktop app, deterministic CLI, and project format that turns layered PSD character art into rigged, animated desktop puppets, with agents driving revisions through a public CLI. It builds on the See-Through layer-decomposition research and never fabricates missing poses.
Hermes Gadget SDK
An ESP32 SDK to build voice and text devices that connect to your own Hermes Agent, including firmware, a Linux client, and a desktop simulator. Notable for making agent hardware accessible via browser-based installation and offering a low-code path to add custom sensors and behaviors.
Project Cascadia
Ports mainline Linux 6.12 to Apple iPad mini 1 hardware using the checkm8 exploit to boot XFCE with touchscreen support. The first public effort to run upstream Linux on Apple's locked A5 silicon.
Ivan Bohun Website
A personal website hosted across six ESP32 microcontrollers without cloud or CDN infrastructure. The blades elect a virtual-MAC leader for TCP load balancing and failover, demonstrating sophisticated distributed systems design on minimal hardware.
Capicola DSP Firmware
Real-time DSP firmware for the Hermetic Modular Alchemy Lab that performs time stretching, pitch shifting, transient detection, and envelope following. It implements a peer-reviewed DAFx26 algorithm and targets hands-on audio experimentation on embedded hardware.
PHP on ESP32
Runs a native PHP 8.3–8.5 interpreter on ESP32 microcontrollers, executing scripts directly from microSD without emulation. A rare bare-metal language runtime that remains actively maintained with a public roadmap.
TinyDesk
A windowed desktop environment for microcontrollers rendered entirely within a terminal, supporting ESP32 boards as well as Windows and Linux hosts. Combines a file manager, editor, and industrial protocols like MQTT and Modbus in a unique TUI interface.
Edge Agents Runtime
A 30 MB runtime for deploying offline AI agents on Linux edge devices like Raspberry Pi and Jetson, with native integration for GPIO, UART, and MQTT protocols. Designed for industrial and IoT use cases where cloud independence is required.
Base Models Can Reason By Taking a Cue From Training Data
Paper reports two opening tokens can bring a base model's reasoning performance close to that of a tuned model.
Claude Code in the cloud: a field guide to cloud sessions
Addy Osmani's field guide to running Claude Code on its own machine: what changes, which workflows pay off, and how to connect GitHub on the first try.