can.ac

Note there is a fork oh-my-pi: https://github.com/can1357/oh-my-pi of https://blog.can.ac/2026/02/12/the-harness-problem/ fame. I use it as a daily driver but I also love pi.

https://blog.can.ac/2026/02/12/the-harness-problem/ It's worse than I expected.

Also vision can be used for "compaction" https://blog.can.ac/2026/06/10/snapcompact/

For the edit tool, you should consider implementing a hash-based approach where each line of code is hashed and referenced by it when doing replacements. You can read up on the approach here: https://blog.can.ac/2026/02/12/the-harness-problem/ I didn't do much benchmarking, but anecdotally, I found it to be making less edit errors. YMMV

Prior art: https://blog.can.ac/2026/02/12/the-harness-problem/

There will always be a gap, but what's interesting is that because new models are constantly coming out, we as an industry never spend any time extracting the maximal value out of an existing model. What if there are techniques, and harness workflows that could be optimized for a singular model end to end? How far can that push the state of the art. An example is https://blog.can.ac/2026/02/12/the-harness-problem/ for just improving edits. Or if we could really steer these open source models using well structured plans, could we spend more time planning into a specific way and kick off the build over night (a la the night shift https://jamon.dev/night-shift)

For the hash-anchored edits, sharing here Can Bölük's original post about the idea https://blog.can.ac/2026/02/12/the-harness-problem/

> it just couldnt write the changes to file 9/10 times when using it with opencode You might want to give this a try, it dramatically improves Edit tool accuracy without changing the model: https://blog.can.ac/2026/02/12/the-harness-problem/

2024 and PatchGuard is still a meme: https://blog.can.ac/2024/06/28/pgc-garbage-collecting-patchguard/

> Fun observation - almost every coding harness (claude code, cursor, codex) uses a find/replace tool as the primary way of interacting with code. This requires the agent to fully type out the code it's trying to edit, including several lines of context around the edit. This is really inefficient, token wise! Why does it work this way? Because the LLMs are really bad at counting lines, or using other ways of describing a unique location in the file. Incidentally, I saw an interesting article about exactly this subject a little ways back, using line numbers + hashes instead of typing out the full search/replace, writing patches, or doing a DSL, and it seemed to have really good success: https://blog.can.ac/2026/02/12/the-harness-problem/

On x86. The linked article loaded very, very slow. Original (?) bloppost: https://blog.can.ac/2021/03/22/speculating-x86-64-isa-with-one-weird-t…

Also see https://blog.can.ac/2021/03/22/speculating-x86-64-isa-with-one-weird-t…