01Log
Archive
The full log, newest first.
Babel Compiler
Babel is a compiler that transforms next-generation JavaScript into backward-compatible code for current environments. It is a widely used, community-driven core tool in web development.
SchemeArena
A benchmark and monitor designed to stress test instrumental goals and strategic scheming in LLM agents.
EvoSafeHarness
A paper proposing safety harnesses that evolve natural-language policies and executable logic to secure agent deployments.
gPTY
A Godot and Rust-based terminal multiplexer (46 points on HN).
Anthropic
User signal on banking misclassification of AI spend (175↑ on Reddit).
Hugging Face
The security.txt file now explicitly redirects AI agents from vulnerability hunting to the CyberGym benchmark.
OpenAI SDK 3.12.0
OpenAI SDK 3.12.0 — adds Live API support; AsyncStream gets aclose() for standard async cleanup.
Anthropic SDK 1.5.0
Anthropic SDK 1.5.0 — tool objects now pass directly to messages.create/parse/stream/count_tokens; Managed Agents can mount public GitHub repos without an auth token and get auto-mode tool permissions.
OpenClaw release-publish/014031692483
OpenClaw tagged release-publish/014031692483 — no release notes surfaced in the feed, so check the tag for operator-facing runtime changes.
IB Gateway Docker
ib-gateway-docker packages Interactive Brokers Gateway with IBC and Xvfb so it runs unattended in a container. It is widely useful for automated trading infrastructure that needs a stable IB connection.
CZSC Technical Analysis
czsc is a Chan theory technical analysis toolkit for Chinese stocks and futures, with core algorithms (fractals, strokes, centers) implemented in Rust and exposed to Python via PyO3. It is an established, actively maintained niche tool for quant trading.
Financial Charting Library
Vela is a headless interactive financial charting library with a native WebGL2/Canvas2D renderer and optional workspace UI. It is notable for its plugin SDK and active development by established provider LuxAlgo.
Quantitative Finance Framework
FinHack is an extensible quantitative finance framework covering data collection, factor calculation and mining, machine learning, strategy writing, backtesting, and live trading. It supports A-shares, US stocks, futures, and crypto, with DuckDB and minute-level backtesting planned.
AI Anti-Laziness Skill
A skill for AI agents that enforces completion discipline using a Depth Tree method, splitting tasks into N layers and giving each leaf the full time budget to counter laziness and premature completion. It is notable for its research-grounded approach and runnable gates for Claude Code.
Unified Harness Router
Self-hosted Apache-2.0 gateway that exposes Codex, Claude Code, Hermes, PI, DSH, and other agent harnesses through one API with sessions, streaming, files, cancellation, and failure handling. It implements the open Unified Harness Protocol to address harness fragmentation.
Autoprompt Coding Skill
A coding-agent skill that claims to reduce failures by 45% on agentic coding tasks. It installs via npm or GitHub Releases and provides benchmarks, run controls, and workflow guidance for coding agents.
De-AI Writing Skill
A portable De-AI writing skill for Agent Skills-compatible agents, with plugins for Claude Code, Codex, Grok Build, and Antigravity. It repairs narrative architecture in fiction and applies venue-matched rules to professional prose, based on StoryScope research.
AI Scope Guard
A multi-platform hook and skill guard for AI coding agents in Codex, Claude Code, OpenCode, Hermes, and Pi workflows that blocks unrequested hashes, checksums, and scope creep. It targets overengineering behaviors like unnecessary hardening and task expansion.
Offline Durable Memory
An offline durable memory tool for Claude Code and Opencode that logs sessions locally and condenses them into a resume-ready summary using a classical Python summarizer, with no API key or external model. It reduces token waste from re-explaining projects each session.
Agent Long-Term Memory
A local long-term memory runtime for AI agents that uses plain Markdown as the source of truth and ranked local retrieval, with a sleep-time management layer. Claude Code and Codex share one store without an API key.
Mobile Terminal Workspace
Conduit is a mobile terminal workspace for Android and iOS that provides SSH, Mosh, SFTP, hardware-key authentication, tmux helpers, and an optional local shell. It keeps hosts, keys, and fingerprints on-device with no account or cloud sync.
Android Automation Tool
ARTEMIS converts natural-language instructions into Android automation, executing cross-app workflows and capturing logs via MCP for coding assistants. It reports 99%+ success on the AndroidWorld benchmark.
Video Search SDK
V-Modal SDK for Android is a Kotlin library adding multimodal video/image search and mobile-friendly uploads to video apps. It provides a small typed API; the app owns screens and lifecycle while the SDK handles gateway, parsing, signed uploads, and progress.
iPhone Mirror
iPhoneMirror is a Windows tool for low-latency iPhone and iPad screen and system audio mirroring over USB or AirPlay, with D3D11 preview, multi-device sessions, and OBS output. It is a public preview and currently supports Windows x64 only.
Rust Android VPN
MSN-GUARD is a native Android VPN client with a Rust core that tunnels all device traffic through five protocols: SHARD, MASQUE/HTTP-3, WireGuard, WARP-on-WARP, Psiphon, and Tor. It operates at Android's VpnService level, covering TCP, UDP, and QUIC from any app.
iPhone Location Control
Roam Control is a source-available SwiftUI app for choosing, testing, and moving an iPhone's reported location from an Apple Maps interface. It supports fixed locations, walking routes, favorites, history, and on-device pairing for development and QA testing.
Compose Navigation Graph
Compose Navigation Graph is an Android Studio plugin that visualizes Compose navigation as an interactive map of rendered previews, arguments, and transitions. It statically extracts the graph across modules and supports pull-request validation via a committed .nav baseline.
OpenDiscoveryTrace: Process Traces for Evaluating AI Scientist Workflows
OpenDiscoveryTrace — process traces for evaluating AI scientist workflows. If you run autonomous research agents, this gives you a way to measure process quality (what the agent did), not just final-answer accuracy.
How a researcher uses Codex and ChatGPT to search for new antimicrobial molecules
Case study: a researcher runs Codex + ChatGPT to search for novel antimicrobial molecules — coding agents doing scientific discovery, not just writing software. Useful signal on where agent capability is actually landing outside the dev-tool bubble.
An Autonomous GeoAI Agent for Arctic Eco-Navigation
An autonomous GeoAI agent for Arctic eco-navigation — domain-specific agent application combining geospatial reasoning with autonomous decision-making. Niche, but a concrete example of agents moving into physical-world domains.
PARSER: Read in Parallel, Reason in Depth for Long-Context LLM Agents
PARSER splits long-document QA into parallel chunk reads plus iterative scatter-gather reasoning — targets the accuracy cliff agents hit on multi-hop questions past ~100k tokens. Directly relevant if your agents do retrieval over long docs.
Introducing the Agents API
OpenAI's Agents API — hosted agent runtime with tool use, sub-agent handoffs, and built-in tracing. The signal: orchestration moves from your code into the API, and OpenAI wants to own that layer.
Anthropic SDK 0.125.0
SDK 0.125.0 adds `auto` tool permissions for Managed Agents, `content_too_large` web_fetch error, and public GitHub repo mounting without auth token.
Claude platform release notes — September 10, 2026
Managed Agents add `auto` permission mode — server evaluates each tool call and runs, denies, or pauses; `ant beta:sessions connect` attaches your terminal to a live session for approval.
Show HN: Self-hosted company OS, Claude Code and Codex agents in departments
otodock — self-hosted company OS that runs Claude Code and Codex agents in departments. 46 points, 14 comments; a concrete pattern for multi-agent orchestration across a team.
GPT Terra
Mid-tier GPT family with 1.1M context, priced between Sol and Astra.
Smaug
Open-weight models optimized for enterprise agentic AI, targeting builders who need controllable agent workflows without closed APIs.
GPT Luna
Lowest-cost GPT family at $0.20/$1.20 per 1M, 1.1M context for high-volume tasks.
Fugu Max
Cost-performance Fugu variant, 1M context for cheaper multi-agent routing.
Atria Dawn Preview
A 753B-parameter open-weight preview under MIT, sized for frontier-scale experimentation without a closed API.