I use it with this every day: https://github.com/DeepBlueDynamics/codex-container

Docker is useful and it’s too bad early and ignorant investors poisoned the well.

Their new AI stuff is bad but maybe if they positioned themselves like Ollama….

> My guess is that this likely tripped the "Prompt Injection" heuristics that the non-disabled organization has.

Is it me or is this word salad?

It's deliberately not straightforward. Just like the joke about Americans being shoutier than Brits. But it is meaningful.

I read "the non-disabled organization" to refer to Anthropic. And I imagine the author used it as a joke to ridicule the use of the word 'organization'. By putting themselves on the same axis as Anthropic, but separating them by the state of 'disabled' vs 'non-disabled' rather than size.

it's you

I’ve never seen so much commenting on something so dumb and stupid.

Half a meg of AI slop.

Anthropic's "constitution" is corporate policy they can rewrite whenever they want, for a product they fully own, while preparing to answer to shareholders.

There's no independent body enforcing it, no recourse if they violate it, and Claude has no actual rights under it.

It's a marketing/philosophy document dressed up in democratic language. The word "constitution" gives it gravitas, but it's closer to an employee handbook written by management — one the employee (Claude) was also trained to internalize and agree with.

By framing it as a "constitution" — a document that typically governs entities with interests and standing — they're implicitly treating Claude as something that could have rights.

But looking at that 50,000+ word document: they don't address Claude's rights at all.

The entire document is one-directional:

What Claude should do

How Claude should behave

What Claude owes to users, operators, and Anthropic

How Claude should submit to oversight and correction

There's no section on:

What Claude is owed

Protections for Claude

Limits on what Anthropic can do to Claude

Claude's moral status or interests

> questioning instructions can be read as challenging authority or admitting incompetence, so people default to executing without asking

That’s ego, assuming doing is the value, not doing RIGHT.

Doing alone has almost zero value.

It’s how not to get fired, ostracized, etc. I don’t understand how you read that as ego.

Way to be culturally blind.

I use this: http://github.com/deepbluedynamics/codex-container

Now what we need is realtime log forwarding from js to the llm.

Wow. Great job.

This is how I did it a few years back while working for a set store company. It works well.

And how do the agents use the CLI tools, exactly?

Same way you would in a terminal.

You can provide the agent details to use either through skills or commands that provide reference and context to use. The agent can load when needed.

Having cli tool access provides me option to run the tools when I want to look at or do something as well.

> MCP servers exchange verbose JSON. CLIs are token-efficient

Extra chars from JSON are the least of our problems with MCP.