Reading list
54 articles I've saved on Instapaper this year, newest first.
October 2026
1 article
September 2026
5 articles
arstechnica.com29 SeptHere's what actually happened in OpenAI's Australian gov't server hackWithout a "full set of safeguards," agent accessed "system information and source code."
lukoplt.blogspot.com7 SeptHow to “Disable” GitHub Copilot Harness Agents in Copilot StudioGovern GitHub Copilot harness agents in two layers: limit Copilot Credits by environment and block the final Dataverse save with a plug-in.
vijayasekhar-deepak.medium.com4 SeptHeadroom vs Caveman vs GraphifyAI Coding Optimization Tools
techcommunity.microsoft.com4 SeptIntroducing Radius Canvas: visualize, review and deploy applications in the GitHub Copilot app | Microsoft Community HubSeptember 3, 2026 - AI can generate thousands of lines of code in minutes. Understanding what it changed across your application, and whether it is ready to...
August 2026
3 articlesjamiemaguire.net9 AugMicrosoft Agent Framework: The Complete Series – Jamie Maguire
404media.co4 AugMicrosoft Tells Engineers ‘Tokenmaxxing Is Not What We Are Optimizing For’Microsoft is introducing budget limits for AI use but says it still wants to be an ‘AI-first’ company.
cityam.com2 AugPwC thought leadership reports '100 per cent AI generated'PwC was found to have published four 'thought leadership' reports that show a pattern of irresponsible AI usage, resulting in hallucinated citations.
July 2026
1 article
June 2026
8 articles
forbes.com27 JunWhy ‘Tokenmaxxing’ Is Out And ‘Valuemaxxing’ Is InTokenmaxxing appears to be on its way out as companies struggle with overspending on AI tokens. Now spend needs to be justified with clear outcomes.
microsoft.github.io27 JunModern Agents Have Skills Now — Here’s How They Work in Copilot StudioHow Skills work in modern Copilot Studio agents: instructions and resources loaded on demand for specific scenarios, why you would modularize instructions into Skills, and when to use Skills versus instructions.
towardsdeeplearning.com21 JunI Built a Monster CLAUDE.md, And My Coding Agent Got Scary Good
valeanu.xyz14 JunTo Preview or Not to Preview? Copilot Studio Reality CheckMicrosoft has been steadily evolving Copilot Studio throughout 2026. With the launch of a new visual workflow canvas and an AI-driven agentic orchestrator, Microsoft has fundamentally reimagined the platform. This evolution moves Copilot Studio beyond basic conversational chat, transforming it into a robust hub for 'Work IQ' and autonomous, multi-step business execution. This article offers a […]
mcsweeneys.net13 JunAI Economics for Dummies“Xavier owns an apartment that he rents out at a loss of $1 billion/month. Seeing this success, he decides to make financial commitments to construct $850 bi...
hellomarvisaitoday.com10 JunNetflix engineer Tejas Chopra creates Headroom, an open-source tool that compresses AI prompts to reduce token costs, and estimates it has saved users $700,000Chopra, a Netflix senior engineer, built Headroom to prune redundant tokens from prompts before they reach language models. He estimates that as much as 90% of tokens are redundant. The tool, not an official Netflix project but already used by several Netflix teams and external projects, has been available since January and currently stands at v0.22, with 2,000 GitHub stars and over 120 forks.
devblogs.microsoft.com7 JunHow Microsoft is migrating repositories to GitHub - Azure DevOps BlogFor the past decade, Azure DevOps has powered software development at Microsoft, supporting some of our largest repositories and most complex engineering
May 2026
7 articles
thenewstack.io30 MayOpus 4.8 Made Claude Smarter. Token Discipline Got Urgent.I’m Matt Burns, Chief Content Officer at Insight Media Group. Each week, I round up the most important AI developments,
xda-developers.com30 MayI set up Claude Code the way Anthropic showed at Code w/ Claude, and it'll change your workflowWho better to copy than the team shipping it every day?
tomshardware.com30 MayMystery company accidentally blew $500 million on Claude AI in a single month — failed to put usage limit on licenses for employees | Tom's HardwareA new report claims AI is getting too expensive for big companies
1password.com15 MayWhat we learned using AI agents to refactor a monolith | 1Password1Password used AI agents to help refactor a massive codebase. The project taught us several lessons about where agents are useful and where they fall short at each stage of the process.
levelup.gitconnected.com3 MayThe 4 Lines Every CLAUDE.md NeedsWhat Karpathy diagnosed, what 60,000 developers bookmarked, and why behavioral constraints beat feature checklists
dev.to1 MayIntroducing fulgur: a blazing fast HTML-to-PDF engine in Rust — no browser requiredHow I built fulgur, a Rust HTML-to-PDF engine on top of Blitz + Krilla — 200-page docs in 190ms with 57MB memory, no Chromium required.
blog.pragmaticengineer.com1 MayThe Pulse: token spend breaks budgets – what next?In the past 2-3 months, spending on AI agents has exploded at many tech companies, Details from 15 of them, including the different ways they are coping with this realization.
April 2026
5 articles
amplifying.ai29 AprAmplifying — Coding Agent IntelligenceBenchmarking and Ground Control for developer tool providers.
openrouter.ai29 AprOpus 4.7's New Tokenizer: What It Actually Costs | OpenRouterThe unified interface for every model. Find the best models & prices for your prompts
medium.com26 AprGoogle makes DESIGN.md open source on its way to become a industry standardLast week, I wrote a post titled ‘Design.md as a standard’ and it’s not like I have inside contacts at Google… or do I? Well, I do, but…
anthropic.com23 AprAn update on recent Claude Code quality reportsAnthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.