In my previous post about riding to Cle Elum, I ran into issues with Trix removing the iframe I was trying to add to show my cycling route on Ride With GPS. Here is the commit with the fixes.

There were multiple issues that needed to be solved:

This issue on GitHub helped me get there, but ultimately I needed to find my own path forward that made sense for this site. Claude code also gave me a great start by writing the majority of the trix-extensions.js file, but I had to do a fair amount of manual clean up to get it to the finish line. But it's so nice to have the AI do the boilerplate to begin with, which gives me a great starting point.

The worst part about this fix was having to patch the trix code directly instead of being able to extend it. But since I vendor the trix code in my repo, it was very easy to do. I actually modified the minified trix.js file directly by ctrl+f "iframe" to remove it from the disallow list. If I need to update trix.js in the future, I'll have to remember to re-apply this fix, so this is definitely a tradeoff, but at least it was an easy way to validate the approach.

I have some lofty cycling goals for this summer, including riding the Michelson Trail in it's entirety. Before I take it on, I need to do build up some endurance. One ride I plan to do which has been on my list for a while is riding to Cle Elum and back, along the SVT and Palouse to Cascades trails.

Edit: This post was supposed to be an excuse for me to try to embed an iframe of the route directly in the post, but I was hamstrung by Trix and DOMPurify. I'm giving up for now!
Edit2: Got it working! I had to add a custom button so that I can paste in a "content attachment" and also modify the dompurify config and trix.js bundle to allow iframes. Lots of hoops to jump through. Here is the commit that got this working.

I ran into an issue using llm today where I was unable to save a response to a file using a pipe

llm llm logs -n 1 | Out-File response.txt

This would give me the error "UnicodeEncodeError: 'charmap' codec can't encode character '\u2192' in position 2831: character maps to <undefined>"

If you set the "PYTHONIOENCODING" environment variable to "utf8", it will fix the issue. This is because Python's default encoding is ASCII. Since the last response I got back from the model contained a non-ASCII character, this error was thrown.

So now, in my PowerShell profile, I've added a line to set the default to utf8, which fixes the issue.

$env:PYTHONIOENCODING = 'utf8'

Quick one-liner to update all llm plugins using PowerShell:

llm plugins | ConvertFrom-Json | % { llm install -U $_.name }

I just found that there is a new(ish) command for figuring out where a transitive dependency comes from in your dotnet project (starting with dotnet 8.0.4xx)

dotnet nuget why <PROJECT|SOLUTION> <PACKAGE>

If you have a dependency in your project that has a vulnerability, you can use this to figure out which package is bringing it in. For example, System.Net.Http 4.3.0 has a high severity vulnerability. I've found instances where this package is brought into my projects by other packages. It's very handy to be able to trace it with a built-in tool. Before this was available, I would use the dotnet-depends tool, which is a great tool, but a little clunkier than I'd like, and doesn't seem to support central package management. 

Blogs are popping off in 2025! Jeremy Warner runs Slate Creek Solutions in the Black Hills and is a great personal friend of mine. I’m excited to see more people start up their own blogs and own their words.

Simon Willison's LLM tool now supports sharing and re-using prompt templates. This means you can create yaml prompt templates in GitHub and then consume them from anywhere using the syntax llm -t gh:{username}/{template-name}.

I have created my own repo where I will be uploading my prompt templates that I use. My most recent template that I've been getting value out of is "update-docs". I use this prompt/model combination to update documentation in my codebases after I've refactored code or added new functionality. The setup is that I use "files-to-prompt" to build the context of the codebase, including samples, then add a single markdown document that I want to be updated at the end. I've found that asking the AI to do too many things at once ends up with really bad results. I've also been playing around with different models. I haven't come to a conclusion on which is the absolute best for updating documentation, but so far o4-mini has given me better vibes than GPT 4.1.

Here is the one-liner command I use to update each document:

files-to-prompt -c -e cs -e md -e csproj --ignore "bin*" --ignore "obj*" /path/to/code /path/to/samples /path/to/doc.md | llm -t gh:david-jarman/update-docs

You can override the model in the llm call using "-m <model>"

llm -t gh:david-jarman/update-docs -m gemini-2.5-pro-exp-03-25

The next thing I'd like to tackle is creating a fragment provider for this scenario so I don't have to add so many paths to files-to-prompt. It's a bit clunky and I think it would be more elegant to just have a fragment provider that knows about my codebase structure and can bring in the samples and code without me needing to specify it each time.

I'm always excited when I see a new .NET Aspire release. The 9.2 update has lots of small quality-of-life improvements. The most interesting new feature for me is "Automatic database creation support". This means that if you add a Postgres database, it will get automatically created, but the best part is that you can also provide a custom SQL script. This lets you customize the creation. I might use this to seed the database with some fake data, so I don't have to manually create data.

This project is showing really great promise. The team is actively listening to developer feedback and reacting to it quickly.

Great article that outlines some of the attack vectors of the Model Conext Protocol. I’ve been playing around with it recently in Claude Code and by attempting to integrate it into the llm CLI by simonw. 

As with any dependency, it’s good to vet the source before using it. Same is true for mcp servers, which are usually docker containers, npm or python tools.

An in depth guide on getting started with Reinforcement Learning by OpenAI.