Is it safe to use without Claude account being suspended?

[LightningFlexors]

That is the only important question.
If my max subscription account gets supended for using third party tools, i can not safely use it.

[askalf]

Short version: I can't promise you won't be actioned, and I'd be wary of anyone who does — only Anthropic decides how it enforces its terms.

Here's what's actually true about how dario works, so you can weigh the risk yourself:

• It runs locally, on your machine, and authenticates with your own Claude login — the same OAuth credential Claude Code uses. Your token never touches my servers or anyone else's; requests go straight from your computer to Anthropic.
• It doesn't modify your account, billing, or subscription. It sends requests the way the official client does.

What it does do is let non-Claude-Code tools use that subscription. Whether that falls within your plan's terms is Anthropic's call, not mine — please read Anthropic's Consumer Terms and Usage Policy and decide for yourself.

One data point, for what it's worth: @askalf/dario has passed 60,000+ npm downloads over the past year, and in that time I haven't received a single report of an account being actioned over it. Read that cautiously — it's absence of reports, not proof, and downloads aren't the same as people running it on a Max plan every day — but it's the closest thing to real-world signal I can honestly give you.

If "zero risk to my Max account" is your bar — completely fair for a paid plan — the only path I can honestly call safe is Anthropic's official apps and the API. I won't oversell dario to you.

[ipaderi]

tldr; so basically dario act as a proxy to send request to OpenAI Chat/Claude Chat instead of the endpoint of Codex/Claude Code.

So in terms of performance and context, doesn't change anything, right? Even the way how models think? Agent functionality still working as should?

Using Codex/Claude Code endpoints it might have better performance or worst?

Thanks for your time.

[askalf]

Wrote this up properly as a standalone reference: #1017 — Will my account get suspended? What is actually observable, and how to monitor it yourself.

It covers the thing that matters most here and that I did not say clearly enough above: the signal you can actually observe is billing classification, not account standing. It also walks through checking your own billing bucket with
dario — Doctor
─────────────

[INFO] dario v5.4.0
[ OK ] Node v24.3.0
[INFO] Platform win32 x64 (10.0.26200)
[ OK ] Runtime / TLS Bun v1.3.14 — TLS fingerprint matches Claude Code
[WARN] CC binary v2.1.235 at C:\Users\masterm1nd\AppData\Roaming\npm\claude (range: v1.0.0 – v2.1.220)
[ OK ] Template live capture, CC v2.1.235 (9h old) (schema v3)
[INFO] Overhead 6,699 chars system prompt + 27 tool defs (89,648 chars JSON-serialized) injected per non-passthrough request. Cached after first hit; read-cost only on subsequent calls within the 5-minute TTL. Exact token count surfaces as cache_creation_input_tokens on the first response (or run dario doctor --usage).
[ OK ] Template drift cache matches installed CC (v2.1.235)
[ OK ] OAuth healthy (expires in 6h 59m)
[WARN] Pool pool of 1, 1 expired (a pool of one — dario accounts add <alias> to load-balance)
[INFO] Pool routing next: login (max-headroom select; 0/1 healthy)
[WARN] Identity 1/1 pool account drifted from ~/.claude.json (live userID=798b515a…): login (accountUuid) — re-run dario accounts add <alias> to refresh the stored snapshot, or non-Haiku requests on the drifted account(s) will 401
[INFO] Backends 1 configured: openai
[ OK ] Sub-agent installed v5.4.0 at C:\Users\masterm1nd.claude\agents\dario.md
[INFO] Home C:\Users\masterm1nd.dario, and publishes the 8 warnings our monitoring canary has raised over 93 days — including the four that turned out to be probe failures rather than anything Anthropic did.

Leaving this discussion open; the answer above still stands.