Both tools turn on the same idea: configurations are named, list_oss_configs enumerates them, and upload_to_oss selects one by configName, so a client can move a file into the right bucket for the environment it is working in. Everything lives on the upload side — there is no listing, download or delete call in the set, and the file has to exist on the host running the server.
A two-tool upload server for Alibaba Cloud OSS, working from named bucket configurations supplied to it at start.
- File upload to OSS: upload_to_oss takes a local file path, an optional stored file name and target directory, and the configuration name that selects which bucket receives it.
- Configuration discovery: list_oss_configs returns the configurations available, which is what makes the configName on an upload something you can check rather than guess.
An Alibaba Cloud OSS bucket and a named configuration carrying its region, access key id, access key secret, bucket and endpoint, supplied to the server at start rather than per call.
One command plus a key — npx oss-mcp --oss-config='{"default":{"region":"oss-cn-shenzhen","accessKeyId":"YOUR_KEY","accessKeySecret":"YOUR_SECRET","bucket":"YOUR_BUCKET","endpoint":"oss-cn-shenzhen.aliyuncs.com"}}' --stdio, then supply credentials