Hello everyone, I also tried with venv, either way you use venv or not, Tiktoken will be the showstopper. Using OpenBSD is not a popular or usual default choice for AI research and development, and that is the point I was trying to make: that being the most security focused platform, IMHO it is a better platform of choice for AI. Other platforms may let me sail easy and that further drives my paranoia that it's easy and therefore too good to be. I'm also looking at llama.cpp and it seems promising given that there's a native package for it, still exploring that option and will share what I find later. Thank you for the advise.
On Fri, Oct 16, 2026 at 4:57 PM Panagiotis Atmatzidis <[email protected]> wrote: > Hi, > > > On 16 Oct 2026, at 7:15 AM, Tito Mari Francis Escaño < > [email protected]> wrote: > > > > Hi misc, > > I'm researching on creating AI agent(s) in OpenBSD 7.9 using CrewAI and > apparentely banging the wall when I try to install Tiktoken. > > I was able to install rust, uv and py3-pydantic-core to make this much > headway. > > Running "pip install tiktoken" doesn't work and gives out error about > externally-managed-environment. Is anyone working on creating a port of > Tiktoken for OpenBSD? > > I'm not familiar with CrewAI, but in general it’s not a good idea to > use system packages to manage dev environments. > > Did you try using python a virtual environment? > > ``` > vm02$ mkdir ~/temp > vm02$ python3 -m ~/temp/venv > temp/venv vm02$ . temp/venv/bin/activate > (venv) vm02$ ...do work here... > ``` > > > I'm unsure it would be smooth sailing after Tiktoken is installed, and > am wondering if CrewAI is a feasible route moving forward to create AI > agent(s) in OpenBSD, but it sure would be more than cool to do so, and > definitely sensibly more secure environment to run these kinds of > automations on > > > My guess is that it won’t be a smooth sailing. > > OpenBSD is IMO far from an ideal system for what > you’re trying to accomplish. On multiple levels. > > > Would appreciate guidance on these concerns. > > Thank you. > > Kind regards, > > P. >
