[0]: https://github.com/openclaw/openclaw/commit/d378a504ac17eab2...
1. Take the oauth credentials and roll your own agent -- this is NOT allowed
2. Run your agentic application directly in Claude Code -- this IS allowed
When OpenClaw says "Open-Claw style CLI usage", it means literally running OpenClaw in an official Claude Code session. Anthropic has no problems with this, this is compliant with their ToS.
When you use Claude Code's oauth credentials outside of the claude code cli Anthropic will charge you extra usage (API pricing) within your existing subscription.
I agree with GP that this is hard to take seriously.
They don't ban Openclaw prompts, each custom LLM application provides a client application id (this is how e.g. Openrouter can tell you how popular Openclaw is, and which models are used the most).
Anthropic just checks for that.
> This is slightly different from what OpenCode was banned from doing; they were a separate harness grabbing a user’s Claude Code session and pretending to be Claude Code.
> OpenClaw was still using Claude Code as the harness (via claude -p)[0]. I understand why Anthropic is doing this (and they’ve made it clear that building products around claude -p is disallowed) but I fear Conductor will be next.
From what I understand, they still had the Claude Code harness available, but were mostly fully integrated on the pi agent framework, using Claude Code's oauth credentials directly,
They also absolutely blocked OpenClaw system prompts from this path in the prior weeks, based purely on keyword detection. Seems they’ve undone that now.
This was widely reported, and happened to me. You probably can’t reproduce it or see it in docs because they seem to have changed the policy.
But then the Claude Code product manager said:
> This is not intentional, likely an overactive abuse classifier. Looking, and working on clarifying the policy going forward.
The fault here is not with Anthropic. It lies with cowboy coders creating a system that violates a providers terms of service and creating an adverse relationship.
One day you're experimenting just fine. The next, everything breaks.
And I'd gladly use their web containerized agents instead (it would pretty much be the same thing), but we happen to do Apple stuff. So unless we want to dive into relying on ever-changing unreliable toolchains that break every time Apple farts, we're stuck with macOS.
The most recent Anthropic announcement was not that people would be banned for using subscriptions with OpenClaw, but that it would be charged as extra usage. I think the reason this was changed three days after that announcement is that being charged for extra usage meant people would not be banned for using their subscription OAuth tokens directly against the Anthropic API with a third party harness, as they had been before. But rather both that usage, and the more recent claude -p usage both be charged as extra usage.
Release notes and announcements are a well-known agentic anti-pattern.
If you're doing them, you're doing agentic wrong. /s-ish-also-cry
Antropic has the info on their website, emailed all users for each step, and I've seen it on X- I'm sure it's in other places as well.
Though, I don't think that justifies spreading FUD in the opposite direction. I also don't think the comment the GP was replying to contains FUD.
Anthropic was, even to me, “one of the better ones” until recently. They have made many questionable/poor decisions the last 6-8 weeks and people are right to call them out for it, especially when they want our money.
I still VERY occasionally use it (as I'm friggin able to anyway) but it's definitely nowhere near my usage previously. And I refuse to give them money, and besideswhich have no goddamn notion of whether it would even be worth it on the lowest paid tier.
Ah well. The free ride was fun but I knew it had a shelf life.
I will say that Codex high/x-high has consistently performed the best for me, but YMMV
Funny thing (or I just imagined that), when I used ChatGPT for studying, it was quite generous about over usage. When I was just messing around, testing where the guardraila are or trying to get it to generate sexual prose about my siblings to send it to them for laughs, the limits were held much more strictly.
I remember when it went up from 25/3h to 50/3h. And I was like meh, because I've already used it over that limit multiple times.
“You’re biased” in these contexts is often just a weak argument bordering on a personal attack. You’re attacking my credibility with no basis for it rather than arguing in favor of anthropic.