I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

A family of lightweight decision models built on Qwen3.5 that you can train and run locally, inspired by Jev-like architectures.

A community effort to revive Amiga Unix (Amix), Commodore's 1990s System V operating system, with modern support for 68040/68060 hardware, a package manager, and contemporary GNU/BSD tools.

Year of the Amiga Unix Desktop 2026

A continual learning model trained from scratch on resource-constrained hardware, using an 8GB VRAM laptop with batch-1 streaming data.

A collection of three major unsolved biological challenges: demonstrating the laboratory emergence of life from chemical precursors, cryopreserving and recovering intact mammals, and creating an enzyme that can reverse-translate peptide sequences into genetic code.

AX is a platform for declaring and running agentic tasks at scale with sandboxed execution, workspace management, network policies, and model configuration. It enables billions of concurrent agent sessions per cluster with sub-second task resumption and dense resource multiplexing.

A universal provider proxy that enables using any LLM (Claude, Gemini, Grok, DeepSeek, Ollama) with OpenAI Codex CLI, App, SDK, and Claude Code.

Explores how to write effectively when large language models are ubiquitous, identifying common patterns of poor LLM-generated writing, defending intentional writing habits often mistaken for AI-generated text, and sharing concrete strategies for using LLMs productively in the writing process.

Shreya Shankar examines how to write effectively when LLMs are part of the writing process, identifying common patterns of AI-generated writing to avoid, defending intentional writing techniques that aren't inherently "LLM-like," and sharing concrete strategies for using LLMs as writing tools while maintaining substance and clarity.

Presents SoL-Pi, a method for improving coding agents through recursive self-improvement loops that reduces token usage by 44.7-49.0% and API costs by about one third while maintaining performance on code generation tasks.

The author argues against using AI for substantive writing tasks, contending that writing is essential to thinking, AI-generated text contains subtle errors, and unlabeled AI writing is misleading to readers.

Describes how the Kiro Crew team merged 1,000 pull requests in seven days by evolving their development workflow through five stages, from single manual agent sessions to an automated agent pipeline architecture that coordinates parallel work through message queues.

Apodex 1.1 develops AI agents capable of complex, long-horizon tasks by scaling both environment diversity (files, search, code execution) and coordination abilities (task decomposition, parallel work, asynchronous integration). The system achieves strong performance across finance, research, mathematics, and coding with a 35B parameter model, grounded in verifiable, real-world work completion.

Documents updates and changes to Claude Code, including new features, improvements, and bug fixes for the AI-powered coding assistant.

Claude now reads AGENTS.md if there is no CLAUDE.md. Finally.

Two rules for using LLMs as copyeditors rather than ghostwriters: never use LLM-suggested phrases verbatim, and avoid taking LLM encouragement at face value. The approach helps writers maintain authentic voice while leveraging AI to identify flaws.

This matches how my own approach to writing has evolved, too.

Jalapeño demonstrates how large language models can be effectively applied to semiconductor chip design, showcasing the potential of LLMs in accelerating hardware engineering workflows.

Skillsync is a local-first tool that makes AI coding sessions portable across different agents and teammates, allowing you to resume work seamlessly without re-explaining context or experiencing vendor lock-in.

LLMs used as classifiers have significant limitations like poor calibration and difficulty incorporating structured data, but treating LLM outputs as features for traditional ML models like logistic regression can overcome these constraints.

A community platform where builders and engineers share their AI tools, workflows, and setups to learn from each other and discover how others work with AI technologies.

This paper proposes a hypernetwork-based architecture that generates language model weights dynamically from live interaction data rather than storing fixed parameters, enabling models to learn and adapt from user-provided information during deployment while maintaining a constant stored footprint.

A curated marketplace of AI agent skills (SKILL.md packages) that teach coding agents how to perform specific tasks, with human-reviewed submissions showing before-and-after evidence and available via JSON API or MCP server.

Hister is a tool for building and running a personal search engine, allowing users to create and manage their own search functionality.

Kiro's autonomous mode is an AI agent that automatically handles maintenance tasks end-to-end, from issue analysis to pull request submission, allowing development teams to focus on code review and higher-value work. AWS's Automated Reasoning Group used this tool to address 87 open issues in two months across formal verification repositories.

Google's document exploring the applications, opportunities, and implications of artificial intelligence across scientific research and discovery.

A technique for fine-tuning the Qwen 4B language model to optimize SQL query plans using reinforcement learning, achieving 81% faster execution than Postgres's default planner.