seed

To automate file intake and seed sources into a processing queue, you invoke the seed command with a file path. The system validates the file presence, calculates the next global claim index, and updates the queue to initiate extraction workflows.

Duplicate detection during file seeding works by comparing both file names and content before processing. If a match is found, the system prevents the duplicate seed, ensuring only unique source files are moved to the dated archive directory and queued for task generation.

Yes, you can seed living documents without relocating them. The system preserves the original file locations for living docs outside the inbox while still generating the corresponding extract task metadata and updating the processing queue.

The best way to ensure deterministic, auditable processing is to seed files through a validation queue. The system records standardized metadata, creates a dated archive directory, and calculates a global claim index to track the extraction workflow.

No, you do not need to manually create an archive directory. The seeding process automatically creates a dated archive directory and moves sources from the inbox into it to keep a permanent record while generating the initial extract task.

Your source file queue needs an extract task artifact to kick off the processing pipeline. Generating this task file updates the queue and establishes the necessary metadata to begin the extraction workflow deterministically.