@feO2x

Engineer Manager at TELIS/GWVS. AI, Cloud-Native and .NET, Speaker, Hobby Shouter and Guitarist @not_brawn

 Regensburg, Germany
Joined November 2009
  • Tweets3.7K
  • Following468
  • Followers337
If you happen to speak German, check out the newest video on @BFreakout's YouTube channel where I was invited to talk all about #GuidedCoding: youtu.be/-jQvlwiCdLw?is=9zRv… #AI #CodingAgents
1
1
64
This is what happens when my wife is away for the weekend with her girls... #AI #CodingAgents
My Claude Code Usage today. Almost 175 million tokens. I would have to pay rougly 900$ if I wouldn't have a Claude Pro Subscription.

ALT My Claude Code Usage today. Almost 175 million tokens. I would have to pay rougly 900$ if I wouldn't have a Claude Pro Subscription.

1
42
Give your #AI #CodingAgent feedback loops. Not only compilers, linters, and automated test suites, but also automated benchmarks. They even write cool reports for you! #csharp #dotnet
A screenshot from claude code, reporting the results of BenchmarkDotNet runs. Overall, performance improved between 16% and 55% in difference scenarios.

ALT A screenshot from claude code, reporting the results of BenchmarkDotNet runs. Overall, performance improved between 16% and 55% in difference scenarios.

1
1
52
Replying to @GearsofWar
@GearsofWar E-Day looks fantastic! Horde Siege seems like a great overhaul - the old 50 wave format was cool, but also grindy. The mission-based approach now looks like it's working great. Can't wait for the beta next week! youtu.be/2-710kEZvK4?si=dad0…
1
35
Replying to @OpenAI
@OpenAI , how sexy is this new jump-to-your-message feature in Codex conversations? Love it! #AI #CodingAgents
1
63
Opus 5 - let's go! #AI #CodingAgents #ClaudeCode
1
1
1
199
Why OpenAI is so much more token-efficient by @theo: youtu.be/ypO0q_8zhWw?si=9SPG…

Why is OpenAI so much more efficient?

OpenAI's models traditionally score higher on coding benchmarks usi...

youtube.com
28
🤗 Today is my lucky day: 5.6 Sol streamed its internal reasoning tokens into my OpenCode conversation. In most cases, we get a summary of the reasoning. OpenAI wants to hide the internal reasoning, mainly so that other labs cannot train their own models on it. #AI #CodingAgents
GPT-5.6 Sol internal thinking tokens were streamed to my OpenCode session. `. It may show "Custom settings layer mounted". Compare. Since user asks informational, could answer directly without tools, but best investigate specific mismatch. We can say yes editorconfig automatic; DotSettings automatic if conventional exact solution shared name. Need likely remove explicit because redundant? User asks maybe update AGENTS? They haven't explicitly asked. They ask question. We shouldn't edit unless clear code expected? Intent informational, no edits. Let's test command verbose and output file. Use bash creates log, no parent issue temp exists verified before. --LogFile writes. No settings vs explicit parallel could write separate. Inspect actual analysis twice maybe okay. Use --verbosity=OFF --LogLevel=TRACE. Then grep logs specialized grep. First run no explicit and second explicit in parallel perhaps cache collision? Avoid same caches perhaps no.

ALT GPT-5.6 Sol internal thinking tokens were streamed to my OpenCode session. `. It may show "Custom settings layer mounted". Compare. Since user asks informational, could answer directly without tools, but best investigate specific mismatch. We can say yes editorconfig automatic; DotSettings automatic if conventional exact solution shared name. Need likely remove explicit because redundant? User asks maybe update AGENTS? They haven't explicitly asked. They ask question. We shouldn't edit unless clear code expected? Intent informational, no edits. Let's test command verbose and output file. Use bash creates log, no parent issue temp exists verified before. --LogFile writes. No settings vs explicit parallel could write separate. Inspect actual analysis twice maybe okay. Use --verbosity=OFF --LogLevel=TRACE. Then grep logs specialized grep. First run no explicit and second explicit in parallel perhaps cache collision? Avoid same caches perhaps no.

2
2
137
In German, we say "Geht runter wie Butter": Thanks to #ADCkonf 2026! I hope to join you again in 2027!
1
31
Step 3: benefit from auto-generated #OpenAPI schemas and examples. In the past, I often only properly documented the happy path in OpenAPI - but now, Light.PortableResults takes care of that!
An image of scalar, a UI for OpenAPI docs, showing auto-generated examples from the Light.PortableResults source generator.

ALT An image of scalar, a UI for OpenAPI docs, showing auto-generated examples from the Light.PortableResults source generator.

1
21
Step 2: in the #aspnetcore Minimal API endpoint where you use the validator, use the new ProducesPortableValidationProblemFor method and specify the validator type to resolve the generic. Use the corresponding attribute in MVC to do the same thing.
C# source code showing how a Light.Portable validator with OpenAPI metadata can be registered with a Minimal API endpoint by calling the ProducesPortableValidationProblemFor<TValidator> extension method.

ALT C# source code showing how a Light.Portable validator with OpenAPI metadata can be registered with a Minimal API endpoint by calling the ProducesPortableValidationProblemFor<TValidator> extension method.

1
16
Step 1: Write a partial validator class and apply the new GeneratePortableValidationOpenApi attribute. This will trigger the source generator which analyzes the validator's PerformValidation method.
C# example code showing a Light.PortableResults validator with the [GeneratePortableValidationOpenApi] attribute applied.

ALT C# example code showing a Light.PortableResults validator with the [GeneratePortableValidationOpenApi] attribute applied.

1
15
The performance benefits are pretty good: for simple flat DTOs, the execution time and memory allocations are roughly 25% less than FluentValidation. For complex DTOs, both numbers go up to over 32%.
Performance benchmarks of FluentValidation vs. Light.PortableResults.Validation. The latter has about 25% less execution time and memory allocations for simple flat DTOs. For complex DTOs, the savings rise to more than 32%.

ALT Performance benchmarks of FluentValidation vs. Light.PortableResults.Validation. The latter has about 25% less execution time and memory allocations for simple flat DTOs. For complex DTOs, the savings rise to more than 32%.

15
How to secure AI agents with Zero Trust! A really helpful overview video from the guys over at IBM. #AI #AiAgents #IBM #ZeroTrust youtu.be/d8d9EZHU7fw?si=HwCC…

Securing AI Agents with Zero Trust

Ready to become a certified SOC Analyst - QRadar SIEM V7.5 Plus Com...

youtube.com
32
Forget Vibe Coding, do Guided Coding! 1) Plan iteratively with the Coding Agent 2) Let the Coding Agent implement the plan 3) Review and iteratively guide the Coding Agent towards better design, security, performance, etc. #AI #CodingAgents #GuidedCoding #VibeCoding
1
1
4
158
Coding agents being able to optimize my code is so cool! In my #dotnet code bases, my agents can use #BenchmarkDotNet to verify that a refactored piece of code is actually faster than before. #AI #OpenAI #CodexApp #csharp
A screenshot of OpenAI's Codex App where the Coding Agent calls into BenchmarkDotNet to verify that a refactoring is faster than the old code.

ALT A screenshot of OpenAI's Codex App where the Coding Agent calls into BenchmarkDotNet to verify that a refactoring is faster than the old code.

1
38
I guess I'm trusting solutions in the ~/.codex/worktrees folder now... #AI #OpenAI #CodexApp #dotnet
The Trusted Locations dialog in JetBrains Rider. I added ~/.codex/worktrees to the list of trusted folders as this is where the OpenAI Codex App places its git worktrees.

ALT The Trusted Locations dialog in JetBrains Rider. I added ~/.codex/worktrees to the list of trusted folders as this is where the OpenAI Codex App places its git worktrees.

37
Load more