דלג לתוכן

Format of an organization's data export

The owner of an organization in LIA can export all its data from Settings › Organizations › «Export the data», also for an archived organization. This page describes what that export contains, file by file, and in which format: it is the format lia.organization-export version 1. Whatever is added later will raise the version, and never change what was already there.

How it is delivered

  • One or more ZIP files, the «parts», of at most 2 GiB each, named lia-organizacion-<id>-<YYYY-MM-DD>-parte-<n>-de-<N>.zip.
  • The first part carries all the data (the JSON files); the uploaded files come after it, in as many parts as needed. An export without files is always a single part.
  • It is prepared in the background, and whoever asked for it is notified when it is ready. It is kept for 7 days; each part is downloaded with a link that lasts 60 minutes.
  • One at a time, and at most a new one every 24 hours. It is free.

The files that describe it

  • README.txt, the first file of the first part: how it is made, in English, with the address of this page.
  • manifest.json, the last file of the last part: the format and its version, the organization, when it was prepared, how it was asked for, whether it carries the files, how many parts it has, every entry of every part (its path, its part, its size, its SHA-256 and, for a list, how many rows it holds) and what could not be included, with the reason (omitted).
  • schemas/: the JSON Schema (draft-07) of each kind of row. Each schema lists every field a row may carry, and no other.

How to read it

  • A .json file holds one object or one list; a .jsonl file holds one row per line (JSON Lines).
  • Keys are in English, those of the LIA API. Dates are milliseconds since January 1, 1970 (UTC); the manifest also gives its own in ISO 8601.
  • People go by their id, and organization/people.json says who each one is. deleted-user is an account that was deleted.
  • Each row goes out as it is stored, but only with the fields of a closed list: anything not on it does not go out. A file copied into the export is named by its zipPath; a link to somewhere else stays a link.

The data, file by file

In the order they are written. Under each one, its fields and its schema.

organization/organization.json

JSON (an object). The organization: its id, name, status, dates and owner.

Fields: id, name, icon, status, ownerUid, createdAt, updatedAt, archivedAt, deletion. Schema: organization.schema.json.

organization/people.json

JSON (a list). Its people: for each member, their id, name, email, role, teams, job title, status and since when they are in it; and those who left.

Fields: uid, name, email, role, teamIds, jobTitle, status, joinedAt, leftAt. Schema: person.schema.json.

organization/teams.json

JSON (a list). Its teams.

Fields: id, name, description, color, emoji, leadUids, status, createdAt, updatedAt. Schema: team.schema.json.

organization/invitations.json

JSON (a list). Its pending invitations, with the invited person's email.

Fields: id, email, role, invitedBy, status, createdAt, expiresAt. Schema: invitation.schema.json.

organization/library.json

JSON (a list). Its Library: what was shared with the organization (with its policy and its teams) and, if it belongs to the organization, the resource itself.

Fields: id, kind, resourceId, name, description, access, status, policy, teams, teamPolicies, requiresPersonalConnection, sharedBy, createdAt. Schema: grant.schema.json.

organization/audit.jsonl

JSON Lines. Its audit log: every row, as the audit log gives it, with its IP.

Fields: id, action, group, at, actor, subject, details, ip. Schema: auditEntry.schema.json.

work/space.json

JSON (an object). Work's settings: its labels, its workflow and its AI settings.

Fields: organizationId, labels, defaultFlow, ai, files. Schema: workSpace.schema.json.

work/projects.jsonl

JSON Lines. Work's projects, with their milestones.

Fields: id, organizationId, key, name, emoji, color, description, mentionUids, itemRefs, objectives, leadUid, teamIds, memberUids, status, health, startDate, targetDate, priority, visibility, flow, estimateUnit, labelIds, links, weeklyClose, weeklyCloseDraft, rollup, archived, createdBy, createdAt, updatedAt, milestones. Schema: workProject.schema.json.

work/updates.jsonl

JSON Lines. The projects' updates.

Fields: id, organizationId, projectId, milestoneId, teamId, author, health, body, mentionUids, itemRefs, createdAt. Schema: workUpdate.schema.json.

work/items.jsonl

JSON Lines. The tasks; for one given to an agent, which agent, who gave it and how its work is going (agent).

Fields: id, organizationId, projectId, key, number, title, description, mentionUids, itemRefs, type, statusId, statusCategory, priority, milestoneId, parentId, createdBy, source, assigneeUids, teamIds, watcherUids, startDate, dueDate, estimate, spent, blocked, blockedBy, blocks, related, duplicateOf, duplicatedBy, labelIds, rank, archived, createdAt, updatedAt, lastActivityAt, startedAt, completedAt, statusChangedAt, commentCount, assetCount, subtaskCount, subtaskDoneCount, agent. Schema: workItem.schema.json.

work/comments.jsonl

JSON Lines. The tasks' comments.

Fields: id, organizationId, projectId, itemId, author, body, mentionUids, itemRefs, createdAt, editedAt. Schema: workComment.schema.json.

work/activity.jsonl

JSON Lines. Work's activity.

Fields: id, organizationId, projectId, itemId, itemKey, level, milestoneId, milestoneName, milestoneNames, type, from, to, fields, subjectUids, actor, at. Schema: workEvent.schema.json.

work/views.jsonl

JSON Lines. The shared views (never the ones somebody did not share).

Fields: id, organizationId, projectId, name, kind, filter, sort, descending, groupBy, lanesBy, properties, portfolio, shared, ownerUid, createdAt, updatedAt. Schema: workView.schema.json.

work/materials.jsonl

JSON Lines. The projects' materials: links and files.

Fields: id, organizationId, projectId, milestoneId, itemId, itemKey, commentId, updateId, type, kind, title, description, url, author, teamIds, labelIds, createdAt, file. Schema: workAsset.schema.json.

work/agent-runs.jsonl

JSON Lines. The agents' runs on tasks: which agent, on which task, who launched it and who stopped it, how it ended and what it spent. Not what it asked to approve nor what it proposed.

Fields: id, organizationId, projectId, itemId, itemKey, agentId, agentName, responsibleUid, trigger, triggeredBy, chatId, state, reason, stoppedBy, credits, attempt, createdAt, startedAt, endedAt. Schema: workAgentRun.schema.json.

work/agent-budgets.jsonl

JSON Lines. Agent spending limits by organization, project and task, including unlimited policies. Revision zero means there has not yet been a human policy edit.

Fields: id, organizationId, version, revision, period, limitCredits, updatedAt, updatedByUid, scope. Schema: workBudgetPolicy.schema.json.

work/agent-budget-changes.jsonl

JSON Lines. Each limit's change history: revision, previous and new amount, date and person who changed it.

Fields: id, organizationId, version, revision, previousLimitCredits, limitCredits, changedAt, changedByUid, scope. Schema: workBudgetChange.schema.json.

work/agent-budget-usage.jsonl

JSON Lines. Spent and reserved credits since tracking began for each scope. This is not an invoice and does not reconstruct earlier spending.

Fields: id, organizationId, version, period, reservedCredits, spentCredits, firstTrackedAt, updatedAt, scope. Schema: workBudgetUsage.schema.json.

work/agent-budget-operations.jsonl

JSON Lines. Model, knowledge and cloud execution reservations with their task, run, state and amounts. Pending reservations stay held until confirmed; this accounting does not create a second charge.

Fields: id, organizationId, projectId, itemId, runId, version, kind, state, quotedCredits, reservedCredits, spentCredits, createdAt, updatedAt, receipt. Schema: workBudgetOperation.schema.json.

work/tool-requests.jsonl

JSON Lines. Metadata for agents' tool requests: task, run, state, dates and approval or rejection decision. The requested content is omitted.

Fields: id, organizationId, projectId, itemId, runId, state, revision, createdAt, updatedAt, expiresAt, contentOmitted, decision. Schema: workToolRequest.schema.json.

work/code-connections.jsonl

JSON Lines.

Fields: id, organizationId, provider, state, revision, connectedByUid, updatedAt, contentOmitted. Schema: workCodeConnection.schema.json.

work/code-repositories.jsonl

JSON Lines.

Fields: id, organizationId, projectId, connectionId, provider, repositoryId, access, enabled, revision, updatedAt, contentOmitted. Schema: workCodeRepository.schema.json.

work/code-task-repositories.jsonl

JSON Lines.

Fields: bindingId, organizationId, projectId, itemId, addedByUid, addedAt, contentOmitted. Schema: workCodeTaskRepository.schema.json.

work/code-pull-requests.jsonl

JSON Lines.

Fields: id, organizationId, projectId, itemId, bindingId, repositoryId, number, state, headCommit, mergeCommit, checks, review, checkedAt, contentOmitted. Schema: workCodePullRequest.schema.json.

work/contexts.jsonl

JSON Lines. Published context for the organization, its teams, projects and tasks, and person context shared with the organization. Drafts and unshared personal context are excluded.

Fields: id, createdAt, scope, published. Schema: workContext.schema.json.

work/memories.jsonl

JSON Lines. Project memories, with provenance, validity dates and review status.

Fields: id, kind, state, title, body, sourceItemId, sourceRunId, proposedByUid, publishedByUid, publishedAt, validFrom, validUntil, supersedesId, revision, createdAt, updatedAt, contentOmitted, scope, sources. Schema: workMemory.schema.json.

work/result-reviews.jsonl

JSON Lines. The acceptance criteria and current review state of each task.

Fields: id, organizationId, projectId, itemId, revision, criteriaRevision, state, updatedAt, criteria, submission, decision. Schema: workResultReview.schema.json.

work/result-submissions.jsonl

JSON Lines. Submitted result versions, with their deliverables and evidence.

Fields: id, organizationId, projectId, itemId, criteriaRevision, summary, submittedAt, criteria, evidence, deliverables, submittedBy. Schema: workResultSubmission.schema.json.

work/result-decisions.jsonl

JSON Lines. Acceptance and change-request decisions and their comments.

Fields: id, organizationId, projectId, itemId, submissionId, outcome, byUid, at, comment. Schema: workResultDecision.schema.json.

work/project-documents.jsonl

JSON Lines. Project document records retained from accepted results, with their origin and state.

Fields: id, organizationId, projectId, itemId, submissionId, state, acceptedByUid, acceptedAt, revision, updatedAt, errorCode, source. Schema: workProjectDocument.schema.json.

company/agents.jsonl

JSON Lines. The organization's agents.

Fields: id, title, description, image, context, _config, _availableFunctions, _dynamicToolDiscovery, _mode, _knowledge, _resources, _apiConnectors, _mcpServers, _integrations, _work, _responsibleUid, _createdAt, _updatedAt, agentConfig, agentTools, handoffs. Schema: agent.schema.json.

company/agent-systems.jsonl

JSON Lines. Its agent systems.

Fields: id, title, description, image, rootAgentId, agentIds, systemIds, _responsibleUid, _createdAt, _updatedAt. Schema: agentSystem.schema.json.

company/chatbots.jsonl

JSON Lines. Its chatbots.

Fields: id, title, description, image, active, maxSessionTimeout, paymentRequired, maxFreeMessages, defaultFreemiumMessage, stripePaymentURL, availableRequestsPerPayment, _responsibleUid, _createdAt, _updatedAt, chat. Schema: chatbot.schema.json.

company/channels.jsonl

JSON Lines. Its channels: their kind, name, chatbots and settings; never how they authenticate.

Fields: id, type, title, description, image, chatbots, rateLimit, dates, allowedDomains, isActive, paymentRequired, maxFreeMessages, defaultFreemiumMessage, stripePaymentURL, availableRequestsPerPayment, freemiumChatbots, stripePaymentLinksAvailableRequests, widget, _responsibleUid, _createdAt, _updatedAt, voiceConfig, email. Schema: channel.schema.json.

company/tasks.jsonl

JSON Lines. Its tasks.

Fields: id, title, description, visibility, status, _executedAt, image, keepChat, _chatId, _responsibleUid, _createdAt, _updatedAt, steps, _cron, _user. Schema: task.schema.json.

company/crons.jsonl

JSON Lines. Its scheduled tasks.

Fields: id, title, description, visibility, status, _executedAt, image, keepChat, _chatId, _responsibleUid, _createdAt, _updatedAt, steps, _cron, _user. Schema: task.schema.json.

company/agentic-tasks.jsonl

JSON Lines. Its agentic tasks.

Fields: id, title, description, agentSystemId, initialMessage, status, _executedAt, image, planId, _disabled, _disabledReason, _disabledAt, timeZone, locale, _chatId, _runId, _responsibleUid, _createdAt, _updatedAt, executions, _cron, _user. Schema: agenticTask.schema.json.

company/agentic-crons.jsonl

JSON Lines. Its scheduled agentic tasks.

Fields: id, title, description, agentSystemId, initialMessage, status, _executedAt, image, planId, _disabled, _disabledReason, _disabledAt, timeZone, locale, _chatId, _runId, _responsibleUid, _createdAt, _updatedAt, executions, _cron, _user. Schema: agenticTask.schema.json.

company/task-executions.jsonl

JSON Lines. The executions of its agentic tasks.

Fields: id, chatId, runId, status, executedAt, errorMessage, _responsibleUid. Schema: taskExecution.schema.json.

company/batches.jsonl

JSON Lines. Its batches. Their data and results go in company/files/batches/.

Fields: id, title, description, image, dataFormat, resultDataFormat, forceJSONResponse, prompt, status, currentRow, totalRows, _chatId, _responsibleUid, _createdAt, _updatedAt, message, data, result. Schema: batch.schema.json.

company/plans.jsonl

JSON Lines. Its plans, with their items, executions and memories.

Fields: id, title, description, strategy, agentSystemId, status, createdBy, createdByAgentSystemId, color, horizon, goals, phases, cron, agenticTaskId, stats, _responsibleUid, _createdAt, _updatedAt, items, executions, memories. Schema: plan.schema.json.

company/api-connectors.jsonl

JSON Lines. Its API connectors: what they do (name, description, operations and the schema of their parameters); never where they call nor with which keys.

Fields: id, name, description, image, method, parametersType, _createdAt, _updatedAt, parameters. Schema: apiConnector.schema.json.

company/mcp-servers.jsonl

JSON Lines. Its MCP servers: what they offer; never where they call nor with which keys.

Fields: id, name, description, image, transport, prefix, readOnlyHintTrust, _createdAt, _updatedAt. Schema: mcpServer.schema.json.

company/integrations.jsonl

JSON Lines. Its connected accounts: the provider, the account, the permissions and the dates; nothing of their access.

Fields: provider, isSandbox, accountId, accountLabel, scopes, subdomain, _createdAt, _updatedAt. Schema: integration.schema.json.

company/triggers.jsonl

JSON Lines. The triggers of those accounts, with their instruction.

Fields: id, provider, event, accountId, title, filters, maxRunsPerDay, agentSystemId, prompt, mailbox, status, createdAt, updatedAt, lastEventAt, lastRunAt, _responsibleUid, target. Schema: trigger.schema.json.

conversations/conversations.jsonl

JSON Lines. Each of the organization's conversations (its channels, chatbots, agents and tasks): its kind, its piece, its title, its dates and the contact as it was saved.

Fields: id, title, description, members, _channelId, _chatbotId, _isAgent, _baseAgentId, _agentSystemId, _createdAt, _updatedAt, lastMessageAt, _humanTakeover. Schema: conversation.schema.json.

conversations/cases.jsonl

JSON Lines. The cases of those conversations.

Fields: chatId, status, since, cycle, openedAt, lastCustomerAt, answeredAt, owedSince, firstReplyAt, lastReplyAt, handledBy, waitingOn, dueAt, resolvedAt, resolution, closedReason, reopened, reopenedAt, firstReplyMs, resolutionMs, updatedAt. Schema: case.schema.json.

conversations/outbox.jsonl

JSON Lines. The replies sent by email: sender, recipient, subject, text and date.

Fields: id, chatId, messageId, status, createdAt, envelope. Schema: outbox.schema.json.

conversations/messages/

JSON Lines. One file per conversation (<conversation>.jsonl): each message, who wrote it (the customer, LIA, deleted-user…), its text, its date and its attachments (their zipPath or their link).

Fields: id, _createdAt, text, system, quickReplies, finishReason, _thinkingContent, user, _knowledgeCitations, _providerReply, attachments, email. Schema: message.schema.json.

knowledge/folders.jsonl

JSON Lines. Its knowledge folders.

Fields: id, name, description, image, icon, tags, documentCount, byteSize, _createdAt, _updatedAt. Schema: knowledgeFolder.schema.json.

knowledge/documents.jsonl

JSON Lines. Their documents: their data (the originals go in knowledge/files/).

Fields: id, folderId, name, kind, sourceType, originUrl, mimeType, byteSize, status, language, media, website, createdAt, updatedAt, indexedAt, zipPath, omitted. Schema: knowledgeDocument.schema.json.

knowledge/pieces.jsonl

JSON Lines. The pieces written by hand in its folders.

Fields: folderId, documentId, id, ordinal, title, body. Schema: knowledgePiece.schema.json.

knowledge/sources.jsonl

JSON Lines. Their sources: what was set up (for a website crawl, its settings; not its pages).

Fields: id, folderId, config, createdAt, updatedAt, lastPublishedAt. Schema: knowledgeSource.schema.json.

The uploaded files

If the export carries the files, they go in these folders, as they were uploaded:

  • work/files/<material>/<name>: Work's files.
  • conversations/files/<conversation>/<message>/<n>-<name>: the attachments of the messages and emails of the organization's conversations.
  • knowledge/files/<folder>/<document>/<name>: the originals and the notes of its knowledge documents.
  • company/files/batches/<batch>/<data|result>-<name>: the batches' data and their results.

Images, logos and voice audio stay as links. An attachment that does not belong to the organization (not from its channels, its Work or its account) is not copied.

Whatever could not be included goes in the manifest's omitted, with its path and the reason:

  • not_found: it was no longer there;
  • unreadable: it could not be read;
  • outside: it does not belong to the organization;
  • knowledge_unavailable: the pieces written by hand did not arrive in time.

What it never includes

Some things are never exported:

  • What is private to each member: their chats and templates in the organization's space (also those of whoever asks for the export), their unshared views, their inbox, their preferences and what they have read, and the drafts of the Friday closes while they are not posted.
  • No secret or credential: not a channel's token, the access keys of WhatsApp, Telegram or Instagram, the addresses, headers and keys of the API connectors and MCP servers, the connected accounts' access, nor their triggers' secrets.
  • What the service only uses inside: the vectors, chunks and extracted text of the knowledge (they are rebuilt from the originals, which do go), Work's search indexes, the upload reservations, the conversations' runs and traces (for an agent working on a task, its summary goes, in work/agent-runs.jsonl), the copies of the charges, the pages of website crawls, and the locks, queues and rate counters.
  • Nothing of another organization nor of anybody's personal space.

If the owner deletes their account

An organization of theirs with nobody else active in it, or an archived one, is kept for 30 days, archived and with nobody able to open it, and is then deleted whole. When deleting their account, its owner gets a link with which, during those 30 days and without an account, they can ask for up to 3 exports in this same format and download them.