The new Claude Desktop app includes a new diff viewer. I compared three ways to review Claude's code changes: the Desktop app diff viewer, the Claude Code CLI, and plain git diff, with a note on when to reach for each one.
www.lotharschulz.info/2026/04/17/c...
#diff #claude_code
Managed Agents APIs
Claude's new Managed Agents API requires a different anthropic-beta header for streaming than for other calls.
www.lotharschulz.info/2026/04/11/c...
#claude #ManagedAgents
Claude Code runs with your full user permissions. nono wraps it in a kernel-level sandbox that makes unauthorised filesystem and network operations structurally impossible.
Claude Code runs with your full user permissions. nono wraps it in a kernel-level sandbox that makes unauthorised filesystem and network operations structurally impossible. Five hands-on scenarios tested on macOS: www.lotharschulz.info/2026/04/11/s...
#claude_code #nono #CLI #AISecurity #sandbox
Gemma 4 Showdown: Linux vs. Mac 🚀
I tested the e4b gemma 4 variant on a 32GB Linux setup vs. a 16GB Mac. The Mac was 4.5x faster (44s vs 199s) and nailed a complex poem constraint. Find more details about _why_ Linux results were different on www.lotharschulz.info/2026/04/04/g... #Gemma4 #AI
🚀 New AI Battle: Gemma 4 on Linux! 🐧
I tested the new Gemma 4 (e4b) running locally via Ollama on Linux. How does it solve the "HORSE-EARTH" poem test?
All technical details: www.lotharschulz.info/2026/04/03/g...
#Gemma4 #Linux #Ollama #OpenSource #AI #MachineLearning #LocalAI #SelfHosted
CDC & Outbox
Build Real Time Data Pipelines with #PostgreSQL and Apache #Nifi: #CDC via logical replication and #Outbox via table polling.
www.lotharschulz.info/2026/01/05/b...
guide including code on building a LLM evaluation pipeline with MLflow
Stop guessing which LLM to use! 🎯
Wrote a guide on building an evaluation pipeline with MLflow. Docker Compose + Python + Ollama = systematic model comparison.
#MLflow #AI #DevOps
www.lotharschulz.info/2025/12/08/c...
Links:
agents guide - github.blog/ai-and-ml/gi...
Long-running agents - www.anthropic.com/engineering/...
Agentic security principles - github.blog/ai-and-ml/gi...
Custom Copilot instructions docs - docs.github.com/en/copilot/h...
"How GitHub’s agentic security principles make our AI agents as secure as possible" -
Clear explanation of how explicitness and boundaries secure Copilot agents. Tip: you may use copilot custom instructions to implement these principles (link below).
"Effective harnesses for long-running agents" - Neat: use two agents (an Initialiser and a Coding agent) to solve the long running agent challenge.
The "Agent failure modes and solutions" section contains great troubleshooting guidance.
"How to write a great agents . md: Lessons from over 2,500 repositories" - Updated this week; includes 5 GitHub Copilot agent proposals. These were great inspiration for the agents I set up this week including a project-specific setup agent (not covered in the blog post).
My top 3 agentic reads this week:
Stop blaming your git skills. This GitHub push failure wasn't about your code — it was about your SSO authority.
A developer could clone a GitHub repository successfully, however failed to push updates to it.
www.lotharschulz.info/2025/11/16/g...
⚠️ #PostgreSQL #CDC danger: Inactive replication slots can accumulate TBs of WAL, crashing your #database!
New post covers:
WAL retention pitfalls
Monitoring best practices
WAL-based CDC vs Outbox Pattern
Don't let silent #WAL growth take you down 🔥
www.lotharschulz.info/2025/10/15/p...
Fixed a prod crash: Snowflake connector down because private key in Azure had no newlines 🤦
Lesson: use `az keyvault secret set --file` for keys. Never copy-paste!
Even whitespace matters in crypto. Full debugging story: www.lotharschulz.info/2025/10/13/w...
#Kubernetes #AzureCloud #Snowflake
Keine #BoundedContext bei Netflix: Irrweg oder Inspiration?
#SoftwareArchitektur im #Stream
HEUTE 13:00 CEST
Mehr Informationen und Möglichkeit für Fragen:
software-architektur...
🧪 Tested Kiro IDE - spec-driven AI development vs traditional coding assistants
Key differentiator: EARS requirements + design blueprints before coding
Great security controls but preview limitations.
www.lotharschulz.info/2025/07/20/k...
#Kiro #DeveloperTools #SpecDriven
It’s like conducting an orchestra where each musician is an AI agent with its own specialty – one handles the frontend, another manages the backend, a third optimizes performance, and they all jam together in perfect harmony.
There’s a new kind of multi-agent development I call “Flow coding”, where you orchestrate a symphony of AI agents and just let them find their groove together. You set the high-level vision, define the boundaries, then step back and watch the magic happen.
Create a weird animals quiz.
I put Claude Sonnet 4 & Google Gemini 2.5 Flash to the test creating "weird animal quizzes."
🚀 Gemini (Canvas): Faster, great look.
⏱️ Claude: Comparable results, but slower.
Both excellent for fast AI prototyping! What are your experiences?
#GoogleGemini #ClaudeAI
Tested Google's new Gemini CLI 🦀
✅ Auto-code generation
⚠️ Sandbox limits external tools
🔧 Highly configurable
I enjoyed security-first vs seamless integration. Shows promise but needs polish.
Full review: www.lotharschulz.info/2025/06/25/g...
#GeminiCLI #AICoding #RustLang
Just had a lightbulb moment 💡
That complex Python script I built 6 months ago for AI-powered supplier due diligence? Modern AI can now do the same thing through simple prompts.
www.lotharschulz.info/2025/06/15/s...
From Jupyter notebooks + API setup → just copy/paste a prompt into AI.
Spent the weekend hacking on a local AI setup for Linux — no cloud, just llm + Ollama.
Wrote a quick guide if you want to try it too 👇
www.lotharschulz.info/2025/06/07/r...
#Linux #AI #LocalAI #OpenSource #WeekendProject #SelfHosted
📢 AI coding tools are creating silent vulnerabilities through "slopsquatting"—where attackers register package names hallucinated by AI.
This attack vector “exploits vibecoding" & targets less technical developers.
#AISecurityRisks #Slopsquatting #VibeCoding
www.lotharschulz.info/2025/05/12/t...
Lost in the software engineering metrics jungle? 🌳 Which framework is right for your team? 🤔
🚀 #DORA 🧠 #SPACEframework 👋 #DevEx 📊 #DXCore4 ⚙️ #ESSP
Find out similarities and differences & how they can transform your software engineering.
Read more: www.lotharschulz.info/2025/05/04/e...
🚀 Bring Rust to AI Workflows!
Learn how to integrate Rust with AI frontends like Claude Desktop using the Model Context Protocol (MCP) by leveraging WebAssembly (Wasm)!
www.lotharschulz.info/2025/04/09/r...
#AI #RustLang #MCP #WebAssembly #AIIntegration #LLMs #ClaudeDesktop
