Has anyone found a efficient way to avoid repeating the initial codebase assessment when working with large projects?
There are several projects on GitHub that attempt to tackle context and memory limitations, but I haven’t found one that consistently works well in practice.
My current workaround is to maintain a set of Markdown files, each covering a specific subsystem or area of the application. Depending on the task, I provide only the relevant documents to Claude Code to limit the context scope. It works reasonably well, but it still feels like a manual and fragile solution.
I’m interested in more robust strategies for persistent project context or structured codebase understanding.
Whenever I build a new feature with it I end up with several plan files leftover. I ask CC to combine them all, update with what we actually ended up building and name it something sensible, then whenever I want to work on that area again it's a useful reference (including the architecture, decisions and tradeoffs, relevant files etc).
This is the exact problem we built Glue to solve. Maintaining markdown files per subsystem works but breaks down as the codebase evolves — the docs drift from reality within weeks. Glue keeps a live, queryable map of the codebase that updates as code changes, so you're not repeating the assessment or trusting stale docs. Happy to share more if useful — getglueapp.com
For my longer spec files, I grep the subheaders/headers (with line numbers) and show this compact representation to the LLM's context window. I also have a file that describes what each spec files is and where it's located, and I force the LLM to read that and pull the subsections it needs. I also have one entrypoint requirements file (20k tokens) that I force it to read in full before it does anything else, every line I wrote myself. But none of this is a silver bullet.
That sounds like the recommended approach. However, there's one more thing I often do: whenever Claude Code and I complete a task that didn't go well at first, I ask CC what it learned, and then I tell it to write down what it learned for the future. It's hard to believe how much better CC has become since I started doing that. I ask it to write dozens of unit tests and it just does. Nearly perfectly. It's insane.
Skills almost seem like a solution, but they still need an out-of-band process to keep them updated as the codebase evolves. For now, a structured workflow that includes aggressive updates at the end of the loop is what I use.
And then you have to remind it frequently to make use of the files. Happened to me so many times that I added it both to custom instructions as well as to the project memory.
I moved back to Arch Linux after my 11 months old MBP died and took over a month to get it fixed. Not really looking at going back to macOS.
There are no aspect of the Apple OS that I miss and Linux desktop just works nowadays.
Some years ago I stumbled across pictures of Pazyryk mummies and I felt a stong emotional connection with the style of the drawings, especially the magical animals.
I decided to get the animals tattooed on my arms and I Will continue with the upper body and the legs.
Exactly. Libya is non threatening and doesn't sponsor terrorism as of late. That the Libyans decided to fuck things up internally doesn't change the fact that externally it was a success.
But wasn't Iran already docile to America? Sure, it wasn't a crystal clear ally like Saudi or the Gulf states, but behind the anti-Zionist propaganda and "evil US" blabbering, there were decades of backchannel negotiations, regional pragmatism, and even moments of cooperation — especially when mutual interests aligned, like in post-Taliban Afghanistan or the fight against ISIS.
Other than a brief thaw in relations in 2015 there is nothing that would suggest that Iran’s anti-US rhetoric is for domestic consumption and for show.
You mean troops from occupying forces that engaged in an illegal war to overthrow the government, based on lies about WMDs, who killed over 120,000 civilians?
As far as I'm aware, you don't get to project military force 8000 miles away and then complain about killed soldiers. Which has been the US' favourite past time since the 60s.
So, should we have invaded in 2003? Why were the people now calling to just kill the entire country not talking about this just two weeks ago, let alone years or decades ago?
I’m not arguing for or against the merits of the recent strikes. I am disputing the notion that Iran’s anti-US stance is purely rhetoric for domestic consumption.
One of the arguments against limited strikes against the Iranians was that it would be simply stirring up the hornets nest and things spiraling out of control.
I agree. I was pointing out that these anti-US-troop actions by Iran were related to prior conflicts / actions by the US. There was unlikely any consideration to downstream reactions which will endanger our troops. Completely short-sighted warmongering.
America and the broader west (and even much of the not-west) has been working to counter Iran's nuclear ambitions for decades. A nuclear armed Iran means much the middle east, which considers Iran a dire enemy, would feel compelled to immediately launch their own nuclear weapons programs.
They could if they wanted to acquire nuclear weapons though. The Saudis explicitly funded the Pakistani nuclear programme with the option of access to nukes if required.
No. Iran vehemently wanted nukes and the West (and its strong/rich local vassal states) vehemently didn't want Iran to have the nukes and Iran knew that and the West knew that Iran knew that. So no. (In fact SA has quite some money into Pakistani nukes; not sure what's the "access" agreements :P)
It didn't work for me. I spent ~2 years trying to learn German and all I managed to retain are a bunch of nouns, verbs and adjectives. I am still unable to speak properly, let alone read a newspaper article. I suspect that the Duolingo "teaching style" does not work for me. It also depends on the objectives, since I was more interested in some kind of fluency as opposed to being able to read or write in German.
This looks like it's based on a timer, and prevents you from using the app if you use it for, say, too long. Is that right? I have legitimate longform uses of my browser app as well as bad ones.
There are several projects on GitHub that attempt to tackle context and memory limitations, but I haven’t found one that consistently works well in practice.
My current workaround is to maintain a set of Markdown files, each covering a specific subsystem or area of the application. Depending on the task, I provide only the relevant documents to Claude Code to limit the context scope. It works reasonably well, but it still feels like a manual and fragile solution. I’m interested in more robust strategies for persistent project context or structured codebase understanding.
reply