A new preprint on arXiv introduces a method for blind watermarking of software repositories by exploiting module placement. The authors argue that many existing source-code watermarking techniques encode a recoverable identifier in tokens or program syntax, which leaves the watermark exposed to discovery or removal.

The abstract, which is truncated, indicates that the new approach diverges from these token- and syntax-based strategies. Instead, it embeds the identifier through the arrangement of modules within the repository, potentially making the watermark harder to detect while still allowing provenance tracking.

Because the full paper is not yet available from the abstract alone, the precise mechanics and robustness claims remain unclear. The work is positioned as a contribution to software provenance, where recoverable identifiers help trace the origin of code artifacts.