pairing endpoint + mode=env

I see https://github.com/soedinglab/MMseqs2-App has an option for passing mode=env to the pairing endpoint but the colabfold config file https://github.com/sokrypton/ColabFold/blob/main/MsaServer/config.json has the environmentalpair config commented out. Looking at the generated pair.sh shell script I would assume this would make the endpoint error but when I hit the pair endpoint with mode=env I get a response. Is there a different environmentalpair db used by the deployed endpoint, or is it not running tip on MMseqs2-App (git history still looks like the environmentalpair config has been there for a while), or am I maybe wrong and MMseqs2-App isn't the right backend repo to look at.

Thanks!

Start with MsaServer/config.json and the generated pair.sh shell script, then compare their environmentalpair settings with the deployed pairing endpoint and the referenced MMseqs2-App repository. Done means establishing which backend and database configuration serves mode=env and recording the confirmed configuration or discrepancy.