Paste a log or an export and keep only the matching lines, or everything except them. Matching runs on contains, starts-with or ends-with.
Filtering is the fastest way to read a file you did not write. Keeping only lines that contain error turns a thousand-line log into the dozen lines that explain the failure. Inverting it, dropping every line that contains a term, is how you strip noise: remove every heartbeat line and what is left is the interesting traffic. Starts-with and ends-with are narrower and more precise. Starts-with matches a log level or a timestamp prefix without hitting the same word inside a message body; ends-with catches file extensions at the end of a path list.
Filters compose well with the other tools here. Filter a log to the error lines, then run the result through the duplicate line remover to see how many distinct errors there actually are, usually far fewer than the raw count suggests. Or filter an export to the rows containing a customer reference, then sort them. Because everything runs locally in your browser, you can do this with a file you would not be comfortable pasting into a server-side tool, which is often exactly the kind of file a log is.
On the whole line, as a substring. A line matches if the term appears anywhere in it, unless you choose starts-with or ends-with.
Not by default, so Error and error both match. Turn on match case when the distinction matters, such as separating a constant from a variable name.
Not in one pass. Run the filter, copy the result back into the input and filter again. Two passes give you the equivalent of an AND.