ramkumareinstein/aid — reverse-engineered prompt

Build me a small command line security tool called AID that scans a folder and finds invisible or suspicious Unicode characters in text files. I want to use it to check code repos or document folders for hidden messages, ASCII smuggling, zero width characters, direction marks, variation selectors, and Unicode tag sequences.

It should skip binary files, scan folders recursively, ignore .git by default, and write a report. The default report should be CSV, but I also want JSON and plain text options. Each report should summarize each affected file with the file path, size, total hidden characters, unique character types, longest run, suspicion level, and any decoded hidden text from Unicode tags.

Please include simple commands like --target, --output, --format, and --verbose. Also add optional switches for scanning classic control characters, suspicious spaces, and Unicode space separators. Make the output easy to understand, with info, medium, high, and critical levels based on how many invisible characters are found and how long the runs are. Use Python and keep it easy to run as a single script.