Improve performance and costs by skipping unchanged files and Git trees during indexing.
- Store Git tree hashes for each branch or tag processed.
- For each file, extract its hash and compare it to previous runs.
- Skip LLM extraction and embedding if content haven't changed.
This avoids redundant processing of identical versions across runs — especially useful when re-indexing updated repositories or processing multiple tags with shared content.
Improve performance and costs by skipping unchanged files and Git trees during indexing.
This avoids redundant processing of identical versions across runs — especially useful when re-indexing updated repositories or processing multiple tags with shared content.