Skip to main content
Your AI tool calls these tools to work with Kadoa. To connect your AI tool first, see Connect your AI tool. The tools you can use depend on your team role. Members, Admins, and Owners can use every tool except approve_workflow, which is reserved for Compliance Officers. Viewers and Compliance Officers otherwise have read-only access. If a tool is not available for your role, you’ll receive an error message with instructions to contact your team admin. The columns below show access for the Member / Admin / Owner, Viewer, and Compliance Officer roles.

Workflows

To find workflows that need action, call list_workflows with statusFilters: ["group:attention"]. It includes workflows where the agent is waiting for your answer, unfinished setup, and previews to approve. Members, Admins, and Owners approve previews in the dashboard. See Preview and approval. status and badges[].label use the same wording as the dashboard. See Workflow lifecycle.

Agent

These tools control Kadoa’s agent. Their names say assistant, but they act on Kadoa’s agent.

Monitors

Runs and extracted data

Change detection

Notifications

configure_notifications takes an events list. Pass ["all"] for every event, or pick from: delete_notification_setting takes a settingId. Earlier versions called this parameter settingsId, so a client or saved prompt that still sends the old name will fail.

Teams and identity

whoami shows the current account, the sign-in method, the active team, and the team’s default product. A connection is fixed to one team. To work in a different team, disconnect and reconnect, and choose that team when you sign in.

Activity

Use list_activity to answer team-wide questions such as who changed a workflow, which workflows failed, or what happened this week. Events are newest first. Use get_workflow_history instead when you need configuration revisions and field-level diffs for one workflow. Filters:
  • Time: startDate and endDate, or relativeTime such as 30m, 6h, 1d, or 2w counted back from now (best for “last week” questions).
  • Who and what: workflowId and userId.
  • eventTypes: CREATED, UPDATED, DELETED, APPROVED, RUNNING, PAUSED, RESUMED, FINISHED, FAILED, DATA_CHANGED, NOTIFICATION_SENT. Workflow run outcomes such as FINISHED and FAILED count as events.
  • interfaces (where the action came from): UI, API, SDK, MCP, CLI, ASSISTANT, SYSTEM, SHELLY, OPS.
  • resourceTypes: WORKFLOW, WORKFLOW_RUN, DATA_QUALITY_RULE, DATA_CHANGE, NOTIFICATION_CONFIGURATION, NOTIFICATION_DELIVERY, SCHEMA, NOTIFICATION_CHANNEL, TEMPLATE, COMPLIANCE, TEAM_MEMBER, INVITATION, TEAM_API_KEY, TAG, TEAM, VARIABLE, USER, AGENT_SESSION, TRIGGER.
  • includeDetails: adds before and after values to each event. Off by default because the result is large.

Observability

get_observability reports a 30-day window by default. Ask for a shorter or longer window when you need it, up to 365 days. It returns a ranked attention list for down and degraded workflows. Add includeWorkflows or includeDailySeries only when you need the larger result.

Variables

Templates

Creating workflows

If your request doesn’t say “run once” or name a schedule, the team’s default product decides. Teams that default to monitoring get create_realtime_monitor. All others get create_workflow. create_workflow takes a prompt, name, URLs, and optionally a schema, plus: create_realtime_monitor needs at least one notification channel. It returns IDs to follow up on with the agent tools.

Updating workflows

update_workflow changes settings directly: name, URLs, schema, prompt (10 to 5,000 characters), schedule, timezone, monitored fields, and row limit.
  • It replaces the whole schema and monitoring value, so your AI tool reads the workflow with get_workflow first.
  • limit: null removes the row limit.
To change what the agent extracts or how it navigates, use request_workflow_update. See Work with the agent.

Reading data

  • fetch_data returns pages of up to 500 rows, 50 by default.
  • export_data returns a signed download link for a whole run, as CSV or JSON. Pass runId for a specific run, or rowIds for up to 1,000 rows.
  • scrape fetches one page right now as Markdown or HTML, without creating a workflow. Page text is capped at 70 KB. It only appears for teams with scraping enabled.
Both data tools take filters, sortBy, and order. A filter looks like {"field": "title", "operator": "CONTAINS", "value": "Manager"}. Several filters are combined with OR.

Large results

List tools return pagination with hasMore and next. Your AI tool keeps calling with next until hasMore is false. fetch_data, get_workflow_history, and get_change stop at about 80 KB, always on a whole row, and set truncated: true. To get the rest, follow next, narrow the filters, or use export_data. get_workflow_history returns field-level changes. Add includeSnapshots: true only when you need full before and after copies, because they are large.

Template versions

update_template only renames a template or changes its description. To change its content, publish a version with create_template_version, then roll it out with apply_template_version. See Templates. create_template_version takes only the parts the version sets: prompt, schemaEntity, a schema (schemaId or inline schemaFields), schemaValidationRules keyed by field name, notifications, and frequency.

Safety and privacy

  • Tools that change or replace configuration, and every delete tool, are marked destructive. Many AI tools ask for your approval before calling them.
  • Notification channel secrets, such as webhook URLs, tokens, and passwords, are masked in results. They never enter your AI tool’s conversation.
  • Results link to the matching page in the dashboard.

Support tickets

create_support_ticket files a ticket for one workflow. Your AI tool should ask you first. If the workflow already has an open ticket, you get that ticket back. Filing a ticket never pauses the workflow.