Academy
CLI and MCP

Dattos CLI: commands and capabilities

Dattos CLI: commands and capabilities

This article is the reference for Dattos CLI commands, organized by area. All commands accept --help for complete flag documentation, and dattos status-glossary explains the meaning of any status returned by the platform. For installation and authentication, see "Dattos CLI: installation and getting started".

Current scope: the CLI covers Process Automation (flows, executions, reconciliation, reports), Task Management, and platform administration. The Account Reconciliation module is not yet available through the CLI — coverage is planned as the product evolves. Requires Dattos 16.4 or later.

Flows and execution (dattos etl)

The CLI's central group — covers the lifecycle of data preparation and reconciliation flows:

  • Create and build with Dattos AI: etl create --name <nome> creates the analysis; etl add-source --etl <id> --file <arquivo> adds sources (xlsx, csv, pdf, xml); etl pipeline create "<descrição em linguagem natural>" --etl <id> builds the complete flow with artificial intelligence (accepts --rules <documento> with the business rules). The etl new shortcut does everything in a single call.

  • Run: etl start <id> --date <AAAA-MM-DD> starts the execution for a reference date; --file uploads the file first, --wait waits until completion, and --incremental preserves data that has already been loaded.

  • Monitor and manage: etl execution-status <id> --date <data>, etl list, etl get <id>, etl resolve <código> (translates a code into identifiers), etl datasources <id>, etl publish <id> (promotes the draft to a published version), and etl activate/etl deactivate.

  • Backup and restore: etl download <id> downloads the flow's .xdtt blueprint; etl upload <id> --file <blueprint> restores it.

  • Calendar: etl set-calendar changes the flow's execution calendar.

  • Advanced building: to build the flow step by step programmatically, use etl add-step, etl connect, etl configure-step, etl set-params, etl add-matching, etl components (catalog and parameters for each step), etl formulas (Dattos Formula reference), etl sql-help, etl step-data (sample output from a step), and etl refresh-sample.

There is no permanent flow deletion. The etl deactivate command is the equivalent of deleting: it hides the flow from listings and can be reversed with etl activate.

Result queries

  • dattos loads: loads get <análise> <data> returns the load and its datasets; loads list-dates lists reference dates with data; loads views lists the available views.

  • dattos datasets: datasets search <análise> <load> <dataset> queries the result rows, with pagination and JSON filters.

  • dattos matching: matching status returns reconciliation totals (reconciled, pending), matching summary provides the summary by column, and matching get returns the records. Rules are managed with matching rules (add, update, delete, active), the catalog with matching configs list, responsible users with matching responsible list, and progress through the approval flow with matching stage.

Export (dattos reports)

  • reports generate <análise> --load-id <id> --dataset-id <id> --format csv|xlsx — asynchronous generation.

  • reports status monitors the report, and reports download <análise> <reportId> --output <arquivo> downloads the result.

Format limit: above 1 million rows, export is available only in CSV.

Task Management (dattos tasks)

Complete read and write access to the task module:

  • List and filter: tasks list combines folder, assignee, label, and due-date filters, with pagination.

  • View details: tasks get <id> returns the full details, including comment and attachment counts.

  • Create: tasks create accepts a title, dates, and recurrence — daily, weekly, monthly, or yearly, with variations such as "weekdays only", "3 business days before the end of the month", or "the second week of the month".

  • Update and move: tasks update changes fields; stage changes always use tasks move (in either direction).

  • Record: tasks comment adds a comment, and tasks attach attaches a file or URL.

  • Export: tasks report --output <arquivo.xlsx> downloads the task report.

Two filters that look similar but are not the same. The stage is the task's manual step (pending, in progress, completed, canceled). The status is calculated by the platform based on due dates (overdue, on time, completed on time, completed late, completed early, canceled).

Administration

  • dattos users: list, create, update, block, and delete users; users set-roles assigns roles; users sessions and users kill-session manage active sessions.

  • dattos roles: management of privacy roles — CRUD, folder scope (roles folders), and reconciliation approval levels (roles approval-levels).

  • dattos connectors: database connectors — create, test (connectors test), and activate/deactivate.

  • dattos calendars: view execution calendars.

  • dattos batch: bulk operations — batch batch-start runs several analyses in sequence; batch batch-create --file <usuarios.json> creates users in bulk.

  • dattos folders: navigation, folder context selection, and folder creation.

Be careful with users set-roles and roles folders set: both replace the entire list rather than append to it. To add an item while preserving the others, use roles folders add (and remove to remove one).

Utilities

  • dattos doctor [--deep] — diagnoses configuration, connectivity, authentication, instance version, and active folder.

  • dattos config profile list|switch — profiles for multiple instances/environments.

  • dattos cache clear — clears the local query cache (5-minute TTL; write operations invalidate it automatically).

  • dattos mcp install <ferramenta> — configures the local MCP server in artificial intelligence assistants. For the hosted option, which is recommended, see the MCP article.

  • dattos version and dattos status-glossary.

Behavior and limits

  • Accepted upload files: .csv, .edi, .ofx, .pdf, .rem, .ret, .txt, .xlsx, .xls, .xlsb, .xml.

  • Deletions require confirmation: destructive commands (users delete, roles delete, connectors delete, matching rules delete, users kill-session) require the --yes flag.

  • Request limit: when it receives HTTP 429, the CLI automatically retries up to 3 times with progressive delays (5s, 10s, 20s).

  • Timeout: 30s for standard calls; 120s for uploads and downloads; 10 minutes for artificial intelligence flows (streaming).

  • Exit codes: 0 success, 1 API error, 2 authentication/configuration — errors are output as JSON to stderr with next_step and correlation_id for support.

Was this helpful?
Was this helpful?