Workspace Export Format

Version 1.0.0, effective from October 4, 2026

This page is the register of the data structures and data formats of the Spyral workspace export. Article 26(b) of the Data Act (Regulation (EU) 2023/2854) asks us to keep this register online and up to date. It describes what a workspace owner receives when exporting the workspace data under section 14.6 of the Terms of Service, so that a firm, or a provider it is moving to, can prepare the import before a switch starts.

How the archive is organised

The workspace export is a single ZIP archive. It contains:

manifest.json, which names the format and its version, describes every data file with its columns, their types and the number of records, and lists what is not included and why;
README.txt, a short guide to the archive;
a data folder with one JSON file and one CSV file for each kind of record (for long text values, such as the text extracted from documents, a JSON file only); and
a files folder with files.json and files.csv, which list every original file with a download link.

Identifiers link the files to each other. For example, company_id in data/company_people refers to id in data/companies, and document_id in data/document_tags refers to id in data/documents, so the records can be linked again in another system. An export never contains data of another workspace.

Encoding

All files are encoded in UTF-8.
Each JSON data file is an array of objects, one object per record. The keys are the column names listed below.
Each CSV file follows RFC 4180: comma-separated, one record per line, lines ending in CRLF, with a header row that names the columns. It starts with a byte order mark so that spreadsheet applications recognise the encoding.
Dates and times are written in ISO 8601 format. Structured values (the jsonb type) are nested JSON in the JSON files and compact JSON text in the CSV files. An empty value is null in JSON and an empty cell in CSV.
In the CSV files, a text value that starts with =, +, -, @, a tab or a carriage return is preceded by an apostrophe, so that a spreadsheet application does not run it as a formula. The JSON files contain the values unchanged.
The column types are the PostgreSQL types of the underlying database columns.

Register

Format identifier	`spyral-workspace-export`
Format version	`1.0`
Encoding	UTF-8. JSON files are arrays of objects; CSV files follow RFC 4180 with a header row.
Download period	7 days after the export is ready, for the archive and the file links

Workspace and people

data/team.json data/team.csvThe workspace (firm) itself: name, plan and eCDF filing identity.16 columns

Column	Type	Can be empty
`id`	serial	No
`name`	varchar(100)	No
`created_at`	timestamp	No
`updated_at`	timestamp	No
`plan_name`	varchar(50)	Yes
`subscription_status`	varchar(20)	Yes
`trial_ends_at`	timestamp	Yes
`onboarding_profile`	jsonb	Yes
`is_sandbox`	boolean	No
`legal_versions`	jsonb	No
`dpa_accepted_by`	integer	Yes
`dpa_accepted_at`	timestamp	Yes
`ecdf_prefix`	varchar(6)	Yes
`ecdf_matricule`	varchar(13)	Yes
`ecdf_rcs_number`	varchar(20)	Yes
`ecdf_vat_number`	varchar(8)	Yes

data/members.json data/members.csvPeople with access to the workspace (name, email, account type).8 columns

Column	Type	Can be empty
`id`	serial	No
`name`	varchar(100)	Yes
`email`	varchar(255)	No
`account_type`	varchar(20)	No
`accepted_terms_at`	timestamp	Yes
`legal_versions`	jsonb	No
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/invitations.json data/invitations.csvInvitations sent to join the workspace.9 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(12)	No
`team_id`	integer	No
`email`	varchar(255)	No
`role`	varchar(50)	No
`invited_by`	integer	No
`invited_at`	timestamp	No
`expires_at`	timestamp	No
`status`	varchar(20)	No

Companies and persons

data/companies.json data/companies.csvClient companies, including soft-deleted ones (deleted_at set).34 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(12)	No
`team_id`	integer	No
`name`	text	No
`rcs_number`	varchar(20)	No
`entity_type`	varchar(60)	Yes
`jurisdiction`	varchar(50)	Yes
`status`	varchar(20)	No
`kind`	varchar(20)	No
`bookkeeping_enabled`	boolean	No
`latest_rcs_extract_id`	integer	Yes
`latest_rcs_extract_date`	date	Yes
`notes`	text	Yes
`is_sample`	boolean	No
`entity_id`	uuid	Yes
`legal_form`	varchar(100)	Yes
`registered_office`	jsonb	Yes
`share_capital`	jsonb	Yes
`custom_fields`	jsonb	Yes
`share_capital_booked_at`	timestamp	Yes
`corporate_purpose`	text	Yes
`nace_code`	varchar(20)	Yes
`incorporation_date`	date	Yes
`financial_year_end`	date	Yes
`matricule`	varchar(13)	Yes
`vat_number`	varchar(8)	Yes
`duration`	varchar(100)	Yes
`auditor`	jsonb	Yes
`associates`	jsonb	Yes
`corporate_events`	jsonb	Yes
`source_document_id`	integer	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No
`deleted_at`	timestamp	Yes

data/company_people.json data/company_people.csvLinks between companies and persons: roles, mandates and shareholdings.15 columns

Column	Type	Can be empty
`id`	serial	No
`company_id`	integer	No
`person_id`	integer	No
`role`	varchar(40)	No
`role_detail`	text	Yes
`shares_held`	numeric(18, 4)	Yes
`share_percentage`	numeric(6, 3)	Yes
`share_class`	varchar(60)	Yes
`mandate_start`	date	Yes
`mandate_end`	date	Yes
`is_active`	boolean	No
`is_primary_signer`	boolean	No
`source_document_id`	integer	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/persons.json data/persons.csvNatural persons known to the firm (directors, shareholders, contacts).14 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(12)	No
`team_id`	integer	No
`first_name`	varchar(200)	Yes
`last_name`	varchar(200)	Yes
`full_name`	varchar(400)	Yes
`date_of_birth`	date	Yes
`place_of_birth`	varchar(200)	Yes
`nationality`	varchar(100)	Yes
`private_address`	jsonb	Yes
`notes`	text	Yes
`manual_override`	boolean	No
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/person_emails.json data/person_emails.csvEmail addresses recorded for persons.7 columns

Column	Type	Can be empty
`id`	serial	No
`person_id`	integer	No
`email`	varchar(255)	No
`label`	varchar(30)	Yes
`is_primary`	boolean	No
`added_by`	integer	Yes
`created_at`	timestamp	No

data/natural_persons.json data/natural_persons.csvNatural persons in the ownership and management graph.7 columns

Column	Type	Can be empty
`id`	uuid	No
`team_id`	integer	No
`full_name`	text	No
`email`	varchar(255)	Yes
`migrated_person_id`	integer	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/ownership_edges.json data/ownership_edges.csvOwnership relations between entities, companies and persons.17 columns

Column	Type	Can be empty
`id`	uuid	No
`team_id`	integer	No
`from_entity_id`	uuid	Yes
`from_person_id`	uuid	Yes
`to_entity_id`	uuid	No
`from_company_id`	integer	Yes
`to_company_id`	integer	Yes
`from_foreign_name`	text	Yes
`from_foreign_rcs`	varchar(50)	Yes
`percentage`	numeric(6, 3)	Yes
`kind`	ownership_kind	No
`effective_from`	date	No
`effective_to`	date	Yes
`source`	ownership_source	No
`source_document_id`	integer	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/management_overlaps.json data/management_overlaps.csvManagement relations between entities, companies and persons.13 columns

Column	Type	Can be empty
`id`	uuid	No
`team_id`	integer	No
`natural_person_id`	uuid	Yes
`manager_entity_id`	uuid	Yes
`entity_id`	uuid	No
`company_id`	integer	Yes
`role`	varchar(100)	No
`effective_from`	date	No
`effective_to`	date	Yes
`source`	ownership_source	No
`source_document_id`	integer	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/relationship_suggestions.json data/relationship_suggestions.csvSuggested relationships extracted from documents, with their review status.11 columns

Column	Type	Can be empty
`id`	uuid	No
`team_id`	integer	No
`client_group_id`	uuid	No
`kind`	suggestion_kind	No
`payload`	jsonb	No
`source_document_id`	integer	No
`confidence`	numeric(4, 3)	No
`status`	suggestion_status	No
`created_at`	timestamp	No
`resolved_at`	timestamp	Yes
`resolved_by`	integer	Yes

Annual accounts

data/client_groups.json data/client_groups.csvAnnual-accounts client groups.6 columns

Column	Type	Can be empty
`id`	uuid	No
`team_id`	integer	No
`name`	text	No
`description`	text	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/entities.json data/entities.csvAnnual-accounts entities.14 columns

Column	Type	Can be empty
`id`	uuid	No
`client_group_id`	uuid	No
`team_id`	integer	No
`rcs_number`	varchar(50)	No
`company_name`	text	No
`legal_form`	varchar(100)	No
`registered_office`	text	No
`share_capital`	numeric(12, 2)	Yes
`financial_year_end`	date	No
`source`	source	No
`shareholder_email`	varchar(255)	Yes
`shareholder_name`	text	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/entity_managers.json data/entity_managers.csvManagers and signers of annual-accounts entities.11 columns

Column	Type	Can be empty
`id`	uuid	No
`entity_id`	uuid	No
`natural_person_id`	uuid	Yes
`manager_entity_id`	uuid	Yes
`full_name`	text	No
`address`	text	No
`organe`	varchar(200)	Yes
`role`	varchar(100)	No
`routing_order`	integer	No
`email`	varchar(255)	No
`created_at`	timestamp	No

data/approval_workflows.json data/approval_workflows.csvAnnual-accounts approval workflows (board and shareholder approvals, filing).21 columns

Column	Type	Can be empty
`id`	uuid	No
`entity_id`	uuid	No
`company_id`	integer	Yes
`financial_year`	integer	No
`status`	workflow_status	No
`result_amount`	numeric(12, 2)	No
`result_type`	result_type	No
`bom_docusign_envelope_id`	varchar(100)	Yes
`shr_docusign_envelope_id`	varchar(100)	Yes
`bom_document_url`	text	Yes
`shr_document_url`	text	Yes
`signed_bom_url`	text	Yes
`signed_shr_url`	text	Yes
`filing_package_url`	text	Yes
`meeting_date`	date	No
`created_at`	timestamp	No
`updated_at`	timestamp	No
`fiscal_year_start`	date	No
`fiscal_year_end`	date	No
`filing_deadline`	date	No
`shr_signer_company_people_id`	integer	Yes

data/signature_events.json data/signature_events.csvE-signature events of approval workflows.7 columns

Column	Type	Can be empty
`id`	uuid	No
`workflow_id`	uuid	No
`signer_name`	text	No
`signer_email`	varchar(255)	No
`signed_at`	timestamp with time zone	No
`docusign_event`	text	No
`envelope_id`	varchar(100)	No

data/approval_generations.json data/approval_generations.csvGenerated approval documents and the field differences they were built from.13 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`company_id`	integer	No
`source_rcs_document_id`	integer	No
`template_document_id`	integer	No
`generated_document_id`	integer	Yes
`approval_kind`	varchar(20)	No
`status`	varchar(20)	No
`field_diff`	jsonb	Yes
`error_message`	text	Yes
`created_by`	integer	No
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/annual_account_documents.json data/annual_account_documents.csvAnnual-accounts source documents and their parsed snapshots.12 columns

Column	Type	Can be empty
`id`	uuid	No
`entity_id`	uuid	No
`company_id`	integer	Yes
`client_group_id`	uuid	No
`document_id`	integer	No
`fiscal_year_start`	date	No
`fiscal_year_end`	date	No
`doc_type`	annual_account_doc_type	No
`parsed_snapshot`	jsonb	No
`source_format`	annual_account_source_format	No
`extraction_confidence`	numeric(4, 3)	No
`created_at`	timestamp	No

Documents

data/documents.json data/documents.csvDocument metadata and classification. The files themselves are listed in files/files.json.32 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(12)	No
`project_id`	integer	Yes
`team_id`	integer	No
`uploaded_by`	integer	No
`filename`	varchar(255)	No
`original_filename`	varchar(255)	No
`file_size`	integer	No
`mime_type`	varchar(100)	No
`storage_path`	text	No
`processing_status`	varchar(50)	No
`processing_error`	text	Yes
`visibility_type`	varchar(20)	No
`allowed_role_id`	integer	Yes
`sync_source`	varchar(30)	Yes
`document_type`	varchar(60)	Yes
`linked_entity_name`	text	Yes
`linked_entity_rcs`	varchar(20)	Yes
`linked_entity_type`	varchar(60)	Yes
`document_date`	date	Yes
`classification_confidence`	varchar(10)	Yes
`classification_source`	varchar(10)	Yes
`linked_company_id`	integer	Yes
`document_family`	varchar(40)	Yes
`document_subtype`	varchar(60)	Yes
`classification_confidence_num`	numeric(4, 3)	Yes
`needs_classification_review`	boolean	No
`bookkeeping_outcome`	varchar(20)	Yes
`bookkeeping_outcome_detail`	varchar(40)	Yes
`is_sample`	boolean	No
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/document_attributes.json data/document_attributes.csvFields extracted from documents.7 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`document_id`	integer	No
`attribute_key`	varchar(60)	No
`attribute_value`	jsonb	Yes
`confidence`	numeric(4, 3)	Yes
`created_at`	timestamp	No

data/document_companies.json data/document_companies.csvWhich companies each document concerns.9 columns

Column	Type	Can be empty
`id`	serial	No
`document_id`	integer	No
`company_id`	integer	No
`team_id`	integer	No
`role`	varchar(32)	No
`confidence`	numeric(4, 3)	No
`source`	varchar(16)	No
`needs_review`	boolean	No
`created_at`	timestamp	No

data/document_counterparties.json data/document_counterparties.csvCounterparties referenced inside documents.8 columns

Column	Type	Can be empty
`id`	uuid	No
`document_id`	integer	No
`referenced_entity_id`	uuid	No
`referenced_company_id`	integer	Yes
`match_type`	counterparty_match	No
`confidence`	numeric(4, 3)	No
`context_snippet`	text	Yes
`created_at`	timestamp	No

data/document_versions.json data/document_versions.csvVersion history of edited documents. The version files are listed in files/files.json.11 columns

Column	Type	Can be empty
`id`	serial	No
`document_id`	integer	No
`branch_id`	integer	Yes
`parent_version_id`	integer	Yes
`version_number`	integer	No
`filename`	varchar(255)	No
`file_size`	integer	No
`storage_path`	text	No
`uploaded_by`	integer	No
`change_description`	text	Yes
`created_at`	timestamp	No

data/document_reviews.json data/document_reviews.csvDocument review requests and decisions.15 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(12)	No
`team_id`	integer	No
`document_id`	integer	No
`version_id`	integer	Yes
`version_number`	integer	No
`requested_by`	integer	No
`assignee_id`	integer	No
`note`	text	Yes
`status`	varchar(20)	No
`decided_by`	integer	Yes
`decided_at`	timestamp	Yes
`decision_note`	text	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/tags.json data/tags.csvLibrary tags.6 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`name`	varchar(100)	No
`color`	varchar(7)	Yes
`created_by`	integer	No
`created_at`	timestamp	No

data/document_tags.json data/document_tags.csvTags applied to documents.4 columns

Column	Type	Can be empty
`id`	serial	No
`document_id`	integer	No
`tag_id`	integer	No
`created_at`	timestamp	No

data/document_text.json JSON onlyText extracted from each document (OCR included). JSON only: values are long.8 columns

Column	Type	Can be empty
`id`	serial	No
`document_id`	integer	No
`text`	text	No
`page_count`	integer	Yes
`source_mime_type`	varchar(100)	No
`extraction_method`	varchar(30)	No
`document_updated_at`	timestamp	No
`extracted_at`	timestamp	No

Bookkeeping

data/bookkeeping_ledgers.json data/bookkeeping_ledgers.csvBookkeeping ledgers and reports, with their journal entries (entries column).28 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(12)	No
`team_id`	integer	No
`project_id`	integer	Yes
`company_id`	integer	Yes
`fiscal_year`	integer	Yes
`ledger_kind`	varchar(20)	No
`template_id`	integer	No
`name`	varchar(255)	No
`description`	text	Yes
`chat_message_id`	varchar(255)	Yes
`user_query`	text	No
`period_start`	date	Yes
`period_end`	date	Yes
`status`	varchar(20)	No
`document_ids`	jsonb	No
`entries`	jsonb	No
`summary`	jsonb	Yes
`output_file_path`	text	Yes
`error_message`	text	Yes
`created_by`	integer	Yes
`created_at`	timestamp	No
`reviewed_by`	integer	Yes
`reviewed_at`	timestamp	Yes
`review_requested_by`	integer	Yes
`review_requested_at`	timestamp	Yes
`review_assignee_id`	integer	Yes
`review_note`	text	Yes

data/bookkeeping_templates.json data/bookkeeping_templates.csvCharts of accounts and bookkeeping templates.15 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(12)	No
`team_id`	integer	No
`name`	varchar(255)	No
`description`	text	Yes
`accounting_standard`	varchar(20)	No
`chart_of_accounts`	jsonb	No
`tax_codes`	jsonb	No
`output_format`	varchar(10)	No
`output_structure`	jsonb	Yes
`default_currency`	varchar(3)	No
`is_default`	boolean	No
`created_by`	integer	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/bookkeeping_entry_links.json data/bookkeeping_entry_links.csvLinks between journal entries and source documents.6 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`report_id`	integer	No
`entry_number`	integer	No
`document_id`	integer	No
`created_at`	timestamp	No

data/bookkeeping_review_events.json data/bookkeeping_review_events.csvBookkeeping review history.9 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`report_id`	integer	No
`type`	varchar(16)	No
`actor_id`	integer	No
`assignee_id`	integer	Yes
`note`	text	Yes
`correction_count`	integer	Yes
`created_at`	timestamp	No

data/bookkeeping_line_reviews.json data/bookkeeping_line_reviews.csvPer-line bookkeeping review assignments.12 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`report_id`	integer	No
`entry_number`	integer	No
`line_index`	integer	No
`assignee_id`	integer	Yes
`requested_by`	integer	No
`requested_at`	timestamp	No
`note`	text	Yes
`reviewed_by`	integer	Yes
`reviewed_at`	timestamp	Yes
`created_at`	timestamp	No

data/bookkeeping_account_memory.json data/bookkeeping_account_memory.csvLearned vendor-to-account mappings.8 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`vendor_name`	varchar(255)	No
`description_pattern`	text	Yes
`matched_account_code`	varchar(20)	No
`match_count`	integer	No
`last_used_at`	timestamp	No
`created_at`	timestamp	No

Projects and tasks

data/projects.json data/projects.csvProjects, including archived and deleted ones.15 columns

Column	Type	Can be empty
`id`	serial	No
`slug`	varchar(12)	No
`team_id`	integer	No
`name`	varchar(200)	No
`description`	text	Yes
`type`	varchar(20)	No
`created_by`	integer	No
`created_at`	timestamp	No
`updated_at`	timestamp	No
`status`	varchar(20)	No
`is_template`	boolean	Yes
`template_id`	integer	Yes
`metadata`	jsonb	Yes
`group_chat_enabled`	boolean	Yes
`deleted_at`	timestamp	Yes

data/project_members.json data/project_members.csvProject membership.6 columns

Column	Type	Can be empty
`id`	serial	No
`project_id`	integer	No
`user_id`	integer	No
`role`	varchar(20)	No
`added_by`	integer	No
`joined_at`	timestamp	No

data/project_companies.json data/project_companies.csvCompanies attached to projects.6 columns

Column	Type	Can be empty
`id`	serial	No
`project_id`	integer	No
`company_id`	integer	No
`team_id`	integer	No
`added_by`	integer	No
`created_at`	timestamp	No

data/project_messages.json data/project_messages.csvProject group chat messages.9 columns

Column	Type	Can be empty
`id`	serial	No
`project_id`	integer	No
`user_id`	integer	No
`content`	text	No
`message_type`	varchar(20)	No
`metadata`	jsonb	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No
`deleted_at`	timestamp	Yes

data/tasks.json data/tasks.csvTasks.13 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(12)	No
`team_id`	integer	No
`project_id`	integer	Yes
`title`	varchar(255)	No
`description`	text	Yes
`status`	varchar(20)	No
`priority`	varchar(20)	No
`assignee_id`	integer	Yes
`due_date`	timestamp	Yes
`created_by`	integer	No
`created_at`	timestamp	No
`updated_at`	timestamp	No

Conversations

data/chat_sessions.json data/chat_sessions.csvAssistant chat sessions of every member, including private ones.10 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	varchar(16)	No
`team_id`	integer	No
`user_id`	integer	No
`project_id`	integer	Yes
`document_id`	integer	Yes
`title`	varchar(255)	Yes
`is_shared`	boolean	No
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/chat_messages.json data/chat_messages.csvMessages of those chat sessions, with the sources cited by the assistant.8 columns

Column	Type	Can be empty
`id`	serial	No
`session_id`	integer	No
`role`	varchar(20)	No
`content`	text	No
`sources`	jsonb	Yes
`mind_map_data`	text	Yes
`model`	varchar(120)	Yes
`created_at`	timestamp	No

data/chat_members.json data/chat_members.csvMembers of shared chat sessions.6 columns

Column	Type	Can be empty
`id`	serial	No
`session_id`	integer	No
`user_id`	integer	No
`invited_by`	integer	No
`joined_at`	timestamp	No
`last_read_at`	timestamp	Yes

Learning and corrections

data/learned_rules.json data/learned_rules.csvRules the workspace learned from reviewed corrections.15 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	uuid	No
`team_id`	integer	No
`surface`	varchar(32)	No
`field`	varchar(48)	No
`context_key`	jsonb	No
`context_key_hash`	varchar(64)	No
`human_value`	jsonb	No
`status`	varchar(16)	No
`occurrence_count`	integer	No
`review_confirmed`	boolean	No
`confidence`	numeric(4, 3)	No
`last_applied_at`	timestamp	Yes
`created_at`	timestamp	No
`updated_at`	timestamp	No

data/corrections.json data/corrections.csvCorrections made by people to AI output.18 columns

Column	Type	Can be empty
`id`	serial	No
`public_id`	uuid	No
`team_id`	integer	No
`user_id`	integer	No
`surface`	varchar(32)	No
`resource_ref`	varchar(128)	No
`document_id`	integer	Yes
`field`	varchar(48)	No
`ai_value`	jsonb	Yes
`human_value`	jsonb	No
`context_key`	jsonb	No
`review_confirmed`	boolean	No
`reason`	text	Yes
`normalized_pattern`	text	Yes
`sharable_consent`	boolean	No
`edit_id`	integer	Yes
`context_key_hash`	varchar(64)	Yes
`created_at`	timestamp	No

data/classification_corrections.json data/classification_corrections.csvManual document re-classifications.11 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`document_id`	integer	No
`user_id`	integer	No
`old_family`	varchar(40)	Yes
`old_subtype`	varchar(60)	Yes
`new_family`	varchar(40)	No
`new_subtype`	varchar(60)	No
`reason`	text	Yes
`origin`	varchar(10)	No
`created_at`	timestamp	No

Activity

data/activity_log.json data/activity_log.csvWorkspace activity log (who did what, when). IP addresses and user agents are left out.5 columns

Column	Type	Can be empty
`id`	serial	No
`team_id`	integer	No
`user_id`	integer	Yes
`action`	text	No
`timestamp`	timestamp	No

Original files

The original files are not copied into the archive. files/files.json and files/files.csv list each one with its kind, its file name, a download link that works for 7 days and the identifiers that link it to the data files. The listing covers:

Original uploaded files.
Files of earlier and edited document versions.
Generated bookkeeping report files.
Annual-accounts approval documents, signed copies and filing packages.

Not included

Data	Reason
Integration credentials (cloud-storage OAuth tokens, Slack/Teams webhook settings)	Secrets; reconnect integrations in the new system.
Embeddings and search vectors	Derived from the documents and the text in data/document_text.json; specific to Spyral's search.
Notifications, bookmarks and saved searches	Per-user state; each member can export it from Settings > Privacy.
Usage metrics, analytics and the cloud-sync queue	Operational telemetry, not content.
Roles and permission policies	Spyral access-control configuration, not portable to another system.

The data that is not exportable because it is specific to the internal functioning of the Services, such as our software, models, prompts and classification rules, is described in section 14.7 of the Terms of Service.

Changes to the format

This register is generated from the definitions the export runs on, so it changes when the export changes: a new kind of record or a new column appears here at the same time as in the archive. When the layout of the archive or of manifest.json changes in a way that an import has to take into account, the format version above changes too. Every archive states in manifest.json the format version it was produced with.

Contact

Questions about the export format, and requests for help with a switch, can be sent to support@spyral.lu. Who we are and how else to reach us is set out in the Legal notice.

Version history

Version	Effective from	Status
1.0.0	October 4, 2026	In force