> If a multimodal model needs to read an image, it cannot use cat for that because the harness needs to inject the actual image payload into the protocol of the LLM.
Sorry, I'm just not familiar, but it sounds like there's a protocol the model expects when receiving images that bash does not support?
(This is Bonteq, I was just logged into the wrong account.)
it cannot use cat but you can make a command which communicates back to agent. Armin mentions that:
> While in theory the agent could provide a CLI tool that talks to the outer harness via environment variables and Unix sockets, it’s a rather crude process
But I don't know why it's crude to be honest. I'm running pi in tmux and have a CLI to prompt it from any shell session / neovim and it works good. So such way of communication is already needed besides codemode.
JS was chosen probably because (1) it's easy to sandbox (there's QuickJS) and (2) (my guess) some models are probably post-trained on JS Codemode.
> But I don't know why it's crude to be honest. I'm running pi in tmux and have a CLI to prompt it from any shell session / neovim and it works good. So such way of communication is already needed besides codemode.
It becomes much crummier when hands and brain are on different machines.
> Agree, but this orthogonal to JS or bash question.
It is not from my perspective because Codemode runs in the brain, and bash necessarily runs where the hands are. So if the hands need to reach into the brain, I need to set up a communication layer from the hands to the brain.
I do this - one profile for work and one for personal. It’s as easy as asking Pi to build it for you. When I run Pi in specific directories it knows which profile to run, using a separate subscription and toolset.
Not natively, as it's still a ToS violation and adding that in pi would go against pi principles, but there are many plugins/proxies that make it work.
oh-my-pi supports it natively (again, still a ToS violation), by impersonating claude code's fingerprints.
I have been using oh-my-pi with 3 claude subs for the past few months without any issues. Even native server-side OAI/ANT compaction works out of the box.
It's pretty unclear because they have two somewhat competing sets of documentation but I believe using the agent sdk with a harness like pi is not against the ToS if it's for yourself.
> Unless previously approved, Anthropic does not allow third party developers to offer claude.ai login or rate limits for their products, including agents built on the Claude Agent SDK.
Matches my experience. Indie games can be delightful, full of Easter eggs and heart that the dev(s) build for the fun of it. I just played through Hollow Knight for the first time and man, what a great experience.
the providers you route to will of course have their own policies, which openrouter surfaces to you through the webui and api. you can even configure automatic routing to select providers based on your policy preferences.
If your data is leaking to OpenRouter, how many companies would be OK with that? So I doubt OpenRouter are peeking inside the workloads that flow through their proxy. However, they do publish "Top Models by Task" ranking[1] so maybe you have a point.
You're devolving into the realm of "What if we tell the agent to just get it right?"
Relying on the prompt to ensure the code it writes is correct is where things fail. Types, tests, linting, etc. are deterministic tools the agents tend to respect.
> If a multimodal model needs to read an image, it cannot use cat for that because the harness needs to inject the actual image payload into the protocol of the LLM.
Sorry, I'm just not familiar, but it sounds like there's a protocol the model expects when receiving images that bash does not support?
(This is Bonteq, I was just logged into the wrong account.)
reply