Batch process scanned I-9 PDFs and reconcile against the active roster.
Pulls I-9 PDFs from the configured OneDrive folder, extracts work authorization data through Claude vision, hashes SSNs for secure matching, and produces a three-sheet audit workbook ready for HRIS import.
Current run
IdlePDFs found
0
Processed
0
Matched
0
Unmatched
0
Low confidence
0
Errors
0
0 of 0
What this run does
- Loads the roster from the configured OneDrive CSV folder, hashes every SSN in memory, discards the raw values.
- Enumerates PDFs recursively through the I-9 OneDrive folder, capturing district and store from the folder path.
- Processes each PDF through Claude vision to extract work authorization status, document details, signature dates, and identifiers.
- Hashes the extracted SSN and matches it against the hashed roster index. Raw SSN never persists; only the last four digits appear in the output.
- Evaluates I-9 timing against the hire date in the roster to flag late completions and pre-hire dates.
- Builds a three-sheet xlsx: Compliance Audit, HRIS Import Data, and Exceptions, available to download on completion.
Security posture
- Cloudflare Access locks every endpoint to a single user account.
- SSN values are SHA-256 hashed with a salted Worker secret. Raw SSN is never written to logs, KV, or the output file.
- Only the last four digits of SSN appear in the audit sheet.
- Job state, per-PDF results, and the roster index expire automatically from KV (2 to 7 days).
- All extraction runs server-side. The browser only sees aggregate counters and the final download link.