ITSpecialist111 finance/invoice-organizer — skill payload
Machine-readable contract:
../invoice-organizer.manifest.json.
What this skill does
Walks a folder of mixed invoice files (PDFs, scanned images), extracts
the supplier / invoice number / invoice date from each via
finance/invoice-extract, then moves each file into a
<supplier>/<YYYY-MM>/ subfolder using the canonical filename
YYYY-MM_<supplier>_<number>.<ext>. Writes a CSV index mapping
original path → new path → extracted fields, so the move is fully
reversible.
When to use it
- A bookkeeper has dumped a month's invoices into one folder and wants them filed.
- You're migrating an AP archive into a structured tree.
When not to use it
- The folder contains anything other than invoices — files that fail extraction are left in place, but for cleanliness pre-filter.
- The target system is a DMS (SharePoint, M-Files) — use its API, don't rename on disk.
Determinism and risk
Determinism: medium — depends on extractor accuracy. Risk: medium — moves files. Mitigations: CSV reversal manifest, dry-run mode, skips files where extraction confidence is below threshold.
How it composes
Hard dependency on invoice.extract (calls finance/invoice-extract
per file). Output CSV is the natural input to a downstream
"reconcile-with-ERP" workflow.