Skip to main content
GET
/
api
/
agents
/
tool-types
Tool types
curl --request GET \
  --url https://api.example.com/api/agents/tool-types
Returns the catalogue of available tool types. Use this to populate tool type selection controls.

​
Response

{
  "status": "success",
  "status_code": 200,
  "data": {
    "items": [
      { "value": "triage", "label": "TRIAGE" },
      { "value": "send_email", "label": "SEND_EMAIL" },
      { "value": "reply_to_email", "label": "REPLY_TO_EMAIL" },
      { "value": "slack", "label": "SLACK" },
      { "value": "trigger_webhook", "label": "TRIGGER_WEBHOOK" },
      { "value": "create_task", "label": "CREATE_TASK" }
    ]
  }
}

​
Tool type reference

ValueDescriptionSandbox mode
triageRead and classify entitiesSafe — executes normally
send_emailSend outbound emails via GmailGuarded — returns would_send
reply_to_emailReply to existing email threadsGuarded — returns would_reply
slackPost messages to Slack channelsGuarded — returns would_send_slack
trigger_webhookFire HTTP webhooks to external systemsGuarded — returns would_trigger_webhook
create_taskCreate work tasks or HIL approval requestsGuarded — returns would_create_task
The “Sandbox mode” column shows how each tool behaves during a sandbox run. Safe tools execute normally; guarded tools return a preview of what the agent would do without actually doing it.