> just ditch

Best solution.

> how can you be sure

For completeness: 

Review the code, import their sha256(seed) in python and test that `sha256(seed).update(<ascii 1-6>).digest()` is deterministic vs, say `@noble/hashes`. Then do a test run on your cold card and note the displayed digest is the same as openssl says for the same sequence.

Take the final outcome from openssl, run it through any bip39 lib, i.e. from the rust crate with the same name. Verify that on a number of derivation paths xpubs, single pubkeys and addresses are the same.

Do all this for testnet and you can also test signing.

optimism

BITC0IN

If you trusted Cold Card to properly use their secure element, how can you be sure they properly used your dice roll entropy? I know you can check the outputs with a CLI script but who actually does that?

I'd go even farther and say just ditch Cold Card altogether. You have nothing to gain by staying and everything to lose.

bitcoin

bitcoin_beginners

PSA: Advice Regarding the Coldcard Bug - Exploit - Hack - Theft

If you trusted Cold Card to properly use their secure element, how can you be sure they properly used your dice roll entropy? I know you can check the outputs with a CLI script but who actually does that?

I'd go even farther and say just ditch Cold Card altogether. You have nothing to gain by staying and everything to lose.