Automate Freshdesk Customer Support

Freshdesk Automation via Rube MCP automates Freshdesk customer-support workflows - ticket management, contact and company operations, notes, replies, and ticket search - through Composio's Freshdesk toolkit. It requires Rube MCP connected (https://rube.app/mcp added as an MCP server, no API keys needed) with an active Freshdesk connection via RUBE_MANAGE_CONNECTIONS, and RUBE_SEARCH_TOOLS must be called first to get current tool schemas before any workflow runs.

Use it across four core workflows: creating/updating/viewing tickets (FRESHDESK_CREATE_TICKET, FRESHDESK_UPDATE_TICKET, FRESHDESK_VIEW_TICKET), searching/filtering tickets (FRESHDESK_GET_TICKETS for simple filters, FRESHDESK_GET_SEARCH for query-syntax search capped at 300 results across 10 pages), replying to and annotating tickets (FRESHDESK_REPLY_TO_TICKET, FRESHDESK_ADD_NOTE_TO_TICKET which defaults to agent-only private notes), and managing contacts/companies (FRESHDESK_SEARCH_CONTACTS, FRESHDESK_CREATE_COMPANIES, etc.). Status, priority, and source are always integers, not strings: status 2=Open/3=Pending/4=Resolved/5=Closed, priority 1=Low through 4=Urgent, source spans 1=Email through 10=Outbound Email.

Ticket creation requires at least one requester identifier (requester_id, email, phone, or a social/external ID) and both subject/description (HTML-formatted) - if phone is given without email, name becomes mandatory too, and attachments must be sent as multipart/form-data, not as file paths or URLs; custom fields must use a cf_ prefix - a gotcha specific to tickets, since company custom fields do not use that prefix. FRESHDESK_CREATE_COMPANIES takes a health_score ('Happy', 'Doing okay', or 'At risk') and account_tier ('Basic', 'Premium', or 'Enterprise') among its fields. Search queries (FRESHDESK_GET_SEARCH) must be double-quoted with single-quoted string values, capped at 512 characters, and support a null keyword for empty-field matches. Contact search requires exact matches (no partial/regex), and company search cannot use the standard name field at all - only custom fields or created_at. ID resolution is a recurring pattern: resolve requester email to requester_id via FRESHDESK_SEARCH_CONTACTS, and company name to company_id via listing and matching, since direct name search isn't supported for companies.

All object IDs (ticket, contact, company, agent, group) are integers with no string-based alternative. Pagination varies by endpoint: FRESHDESK_GET_TICKETS allows up to 100 per page, while search and contact endpoints cap at 30 per page across a maximum of 10 pages. Two reply tools exist (FRESHDESK_REPLY_TO_TICKET with more options like CC/BCC, and the simpler FRESHDESK_REPLY_TICKET) - both work, but only notes (not incoming replies) can be edited afterward via FRESHDESK_UPDATE_CONVERSATIONS. In notes, notify_emails only accepts agent email addresses, not customer/external contacts. Bulk contact import runs through FRESHDESK_IMPORT_CONTACT from CSV, and rate limits apply per-account by plan tier, so bulk operations should be paced to avoid 429 responses.

Support teams automating Freshdesk operations - ticket lifecycle, customer replies, internal notes, and contact/company data management - through an MCP-connected agent rather than direct API integration, with the parameter quirks (integer status codes, prefix differences, exact-match search limits) already documented to avoid common mistakes.