Indexes folder and file names into a SQLite database — one row per file — for querying and for the rest of the WETTER pipeline.
Download: latest release — DAMPF-vX.Y.Z.exe (Windows) / DAMPF-vX.Y.Z-linux-x86_64 (Linux).
- Download the binary for your platform.
- Linux:
chmod +x DAMPF-vX.Y.Z-linux-x86_64, then run it. Windows: run the.exe. - Set the crawl root, click Crawl, adjust which keys to keep, then Save to DB.
Step 4 — Detected keys
- Used — which keys become columns. None checked means all keys; any checked means only those.
- Type — Numeric or Category. A folder or underscore block that changes becomes a column (Category). More than 10 distinct numbers become Numeric. Category stores TEXT so WOLKE shows a filter instead of a histogram. A block that is the same in every file stays gray. The dropdown overrides the suggestion.
- Uniques — distinct values in the current file set.
- Full — how many files have the key;
(−N)means N files are missing it.
Naming: Cheat sheet · full convention
For agents: docs/llm-brief.md.
Developed by M.E.S.S..