Pennylane
Pennylane is accounting and invoicing software for French SMBs.
Operational order
- Company profile once per session — confirms legal entity, currency, and fiscal settings before party or invoice work.
- Customers and suppliers when you need party ids for invoice filters.
- Products and categories when building or checking invoice line references.
- Customer invoices / supplier invoices for AR/AP — always paginate; large firms span many cursor pages.
- Ledger accounts, journals, and ledger entries for GL structure and postings.
- Quotes for open sales quotes before they convert to invoices.
- Bank accounts and transactions before payment matching actions.
- Changelogs for incremental sync — set resource to the entity family; use start_date on first page, then cursor only.
Reads
Omit id to list with cursor pagination; pass id to fetch one row.
Bulk GL extracts
For full-period books, use the export action — not unbounded ledger line pagination. Start with export_type (fec, general_ledger, or analytical_general_ledger), period_start, and period_end; poll with the returned id until status is ready; set download=true on the poll call to fetch via a spillover link (file links expire in ~30 minutes). Analytical exports accept mode in_line or in_column.
Analytical categories
One categorize write covers customers, suppliers, both invoice types, and bank transactions — pass resource plus a JSON array of {id, weight} (weights in the same group must sum to 1).
Pagination
- Use cursor (opaque token from next_cursor) and limit (default 25, max 100 on most lists; changelogs allow up to 1000).
- Re-send filter and sort on every page when the list is filtered or sorted.
- Changelog feeds: use start_date OR cursor, never both. Retention is ~4 weeks.
- Do not request huge pages on invoice lists — filter by customer_id, supplier_id, or date when possible.
Trial balance
Requires period_start and period_end (YYYY-MM-DD) plus optional cursor pagination.
Invoice lifecycle
- From quote — convert an accepted quote into a customer invoice.
- Import — upload the PDF or e-invoice file first (returns file_attachment_id), then import customer or supplier invoice with that id.
- Finalize — draft customer invoices only.
- Email delivery — finalized or imported customer invoices and quotes; if the PDF is still generating, wait a few minutes and try again.
- Match payment — link bank transactions to finalized customer or supplier invoices.
- Validate AP — mark supplier invoice accounting complete.
- Quote status — pending, accepted, denied, invoiced, or expired (separate from line edits).
Filtering
Optional filter is a JSON array: [{"field":"…","operator":"eq","value":"…"}].
Common operators: eq, not_eq, lt, lteq, gt, gteq, in, not_in (field support varies by resource).
Writes
- Upsert pattern — omit id to create; pass id to update. Use body_json when typed fields are not enough.
- Customers — set customer_type to company or individual.
- Supplier invoices — updates require an existing id; importing a file uses the dedicated import action with body_json when needed.
- Delete customer invoice — draft invoices only.
- Monetary amounts are decimal strings, not floats.
- File upload — inline base64 is capped at 1 MB; use file_url for larger PDFs (up to 100MB). Local-only files over 1 MB need MCPBundles Desktop to stage bytes first. A presigned download link from a prior large read can be passed as file_url.
- Nested reads — pass parent_id for invoice lines, payments, matches, customer categories/contacts, commercial document lines, etc.
Large read payloads
List and detail responses return JSON inline when small. Oversize payloads return a presigned download link — open it for full JSON, or pass that URL as file_url on upload/import steps.
Data model
- Customers, Suppliers — parties.
- Customer invoices, Quotes, Supplier invoices — billing documents.
- Products, Categories, Category groups, Ledger accounts — catalog and chart structure.
