registerCompletionItemProvider API from VS Code extension does not to trigger provideCompletionItems

Hey I am trying to register some snippets in VS Code dynamically. Therefore I am going over multiple files and collecting all snippets I find. With the Collected data I call then for every File this function (I guess this is pretty ineffective, however, I first want to solve the issue of why no snippets at all show up):

const registerCompletionItemProvider = (
    snippetLanguage: string | string[],
    snippetObject: any,
    name: string
) => {
    console.log(snippetLanguage);
    console.log(snippetObject);
    console.log(name); // this executed
    vscode.languages.registerCompletionItemProvider(
        // snippetLanguage,
        [{ scheme: 'file', language: 'javascript' }], // put this here for debugging
        {
            provideCompletionItems(document, position) {
                // this never runs
                const start = new vscode.Position(position.line, 0);
                const range = new vscode.Range(start, position);
                const completionItems = Object.keys(snippetObject).map((key) => {
                    const snippet = new vscode.SnippetString(snippetObject[key]);
                    const completionItem = new vscode.CompletionItem(
                        key,
                        vscode.CompletionItemKind.Snippet
                    );
                    completionItem.range = range;
                    completionItem.insertText = snippet;
                    return completionItem;
                });
                return completionItems;
            },
        },
        name
    );
};

However somehow, it seems that the provideCompletionItems is never executed in a .js. I expected that it should trigger when I enter the name of the Snippet or the prefix?

{
    "First": { // First would be one name
        "prefix": [
            "test1" // this would be the prefix 
        ],
        "body": [
            "function ${1:myFunction}($2) {",
            "\t$0",
            "}"
        ],
        "description": "Any Function"
    },
    "Secondary": {
        "prefix": [
            "test"
        ],
        "body": [
            "function ${1:myFunction}($2) {",
            "\t$0",
            "}"
        ],
        "description": "Any Function"
    }
}

This looks to me like a misunderstanding of what a triggerCharacter is, i.e. the third argument of the registerCompletionProvider method.

This has to be a list of single-character strings, usually "." or ":". When the user types one of these, your completion provider would be called, just as if the user typed Ctrl+Space to bring up a list of suggested completions. Your provider then needs to analyze or parse the text around the current position -- In your case something that discovers word boundarys and then provide matching completion items. For example if the position is somewhere on or right after "te", return the snippets for "test1" and "test".