Reading and searching

read_file, grep, and glob are how tCode gets its bearings in a project it has never seen before it changes anything in it. read_file reads a specific file it already knows the path to; grep searches file contents for a pattern across a directory; glob finds files by name pattern when the exact path is not yet known.

A typical investigation uses more than one of these in sequence: a glob to find the right kind of file, followed by a grep inside the results to narrow down which one actually matters, followed by a read_file on the specific one that turned out to be relevant.

None of the three can affect anything outside the search itself, which is exactly why they run without ever interrupting you for approval; there is nothing at stake in looking.

See grep versus glob for the specific difference between the two search tools, and when a given investigation calls for one over the other or both together.

This orientation phase is not something you have to explicitly request; it happens naturally as part of the model working out how to address your message, the same way you would look around a codebase yourself before touching anything in it for the first time.