Friday Roundup - Week 21: Google I/O, GitHub Breach, Agents
Google I/O ships Gemini 3.5 and Jules; GitHub discloses a supply chain breach of 3,800 repos; Claude Code adds agent coordination; Arazzo 1.1 ships.
Blog
Friday Roundup - Week 20: OSS Strip Mining and Mythos
LLM-powered OSS vulnerability scanning crossed a practical threshold, Anthropic gates Mythos, swagger-php 6.1.2 ships, and TinyML reaches smallholder farms.
Greenhouse Climate Modeling with ASHRAE and Python
How cloudgrow-sim turns American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) psychrometrics, solar gain, heat transfer, and ventilation equations into a Python model validated against weather and sensors.
Friday Roundup - Week 19: TypeScript 10x, Agents, and Overlay
TypeScript 7.0 beta ships a Go-rewritten compiler at 10x speed, agentic search challenges the retrieval abstraction, and OpenAPI Overlay v1.1.0 lands with practical toolchain wins.
Why Service Dependency Chains Violate the Reliability Product Law
The CrowdStrike loss exceeded an independence-model prediction by three orders of magnitude. That gap is the empirical signature of common-cause coupling.
When AI Code Errors Compound: A Math Check on Dependency Risk
A YouTuber sketched an equation for AI-induced downtime as service dependencies deepen. The intuition is right, the formula is inverted, and the corrected version still understates the real risk. Here is the math.
Friday Roundup - Week 18: MCP Governance, Security, and AgTech AI Agents
MCP's donation to the Linux Foundation collides with a critical RCE advisory. Claude Code exits the $20 Pro tier. GROWMARK deploys an AI agronomist. Three arXiv papers on recursive agents and benchmark design.
Most Data Contract Tools Don't Enforce Contracts: Here's What Does
OSS data contract tooling in 2026 mostly stops at definition. Four projects actually enforce contracts at runtime. The rest hand you a YAML file and a hope.
Why Enterprise AI Agents Route on Intuition Instead of Knowledge
Enterprise multi-agent systems route work by embedding similarity. Recent research shows formal ontologies outperform that approach by measurable margins.
Friday Roundup - Week 17: Postmortems, Frontier Models, and Planting Season AI
Anthropic's April 23 postmortem on three Claude Code bugs, DeepSeek V4 at 1,256 HN points, MCP architecture RCE exposure, and AI seeding prescriptions entering the 2026 planting season.