Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.svantic.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Zendesk Plugin for Savant

A standalone A2A agent that bridges Zendesk ticket events into the Savant mesh and exposes Zendesk operations as tools for other agents.

​
What It Does

  • Listens to Zendesk webhooks (ticket.created, ticket.updated, ticket.status_changed, ticket.escalated) and polls the Zendesk Incremental API as a fallback
  • Transforms ticket events into structured Savant cues routed to the mesh
  • Exposes 8 Zendesk tools (get ticket, create ticket, update ticket, add comment, search, list, get user, apply macro) callable by any agent on the mesh

​
Architecture

The plugin is built with @svantic/sdk and connects via MeshConnector. It registers as a remote agent on the mesh, visible in the dashboard Agents page.

​
Deployment

ZENDESK_SUBDOMAIN=acme \
ZENDESK_EMAIL=bot@acme.com \
ZENDESK_API_TOKEN=xxxxxxxx \
ZENDESK_WEBHOOK_SECRET=whsec_xxxxxxxx \
SAVANT_MESH_URL=http://savant-mesh:3000 \
SAVANT_CLIENT_ID=your-client-id \
SAVANT_CLIENT_SECRET=your-client-secret \
node dist/index.js

​
settings.json

{
    "plugin": "savant-zendesk-plugin",
    "version": "1.0.0",
    "agent": {
        "name": "savant-zendesk-plugin",
        "description": "Bridges Zendesk ticket events into the Savant mesh.",
        "port": 4100,
        "agent_type": "zendesk-plugin"
    },
    "mesh": {
        "url": "http://localhost:3000",
        "tenant_id": "acme",
        "tenant_secret": "${SAVANT_TENANT_SECRET}"
    },
    "zendesk": {
        "subdomain": "${ZENDESK_SUBDOMAIN}",
        "email": "${ZENDESK_EMAIL}",
        "api_token": "${ZENDESK_API_TOKEN}",
        "webhook_secret": "${ZENDESK_WEBHOOK_SECRET}"
    },
    "event_ingestion": {
        "webhook_path": "/webhooks/zendesk",
        "polling_enabled": true,
        "polling_interval_seconds": 60
    },
    "routing_rules": [
        {
            "match": { "priority": ["urgent", "high"] },
            "action": { "auto_execute": true, "cue_priority": "high" }
        },
        {
            "match": { "tags_include": ["escalated"] },
            "action": { "requires_human": true, "cue_priority": "high" }
        },
        {
            "match": { "event_type": "ticket.status_changed", "status": ["solved", "closed"] },
            "action": { "log_only": true }
        }
    ],
    "tools": {
        "enabled": [
            "zendesk_get_ticket",
            "zendesk_list_tickets",
            "zendesk_create_ticket",
            "zendesk_update_ticket",
            "zendesk_add_comment",
            "zendesk_search",
            "zendesk_get_user",
            "zendesk_macro_apply"
        ]
    }
}

​
Tools

ToolDescription
zendesk_get_ticketFetch full ticket details with comments
zendesk_list_ticketsSearch/list tickets with filters
zendesk_create_ticketCreate a new ticket
zendesk_update_ticketUpdate ticket fields
zendesk_add_commentAdd a public or internal note
zendesk_searchFull-text search across tickets
zendesk_get_userLook up a Zendesk user
zendesk_macro_applyApply a predefined macro

​
End-to-End Example

  1. Customer creates high-priority ticket: “Can’t log in”
  2. Zendesk fires webhook to plugin
  3. Plugin transforms to cue, routes to mesh (auto_execute due to high priority)
  4. Planner generates plan: fetch ticket → query KB → draft response → post comment
  5. Executor runs plan, calling zendesk_get_ticket and zendesk_add_comment on the plugin
  6. Customer sees the response on their ticket within seconds