Integrating Dust with GitHub and Zendesk: A Guide to Streamlined Support Ticketing and Issue Triage

Customer support engineers and technical operations teams routinely face a frustrating operational barrier: customer issues arrive in Zendesk, while code fixes, bug tracking, and developer discussions live in GitHub. When these two environments operate in silos, support agents spend hours manually searching GitHub repositories, copying stack traces, duplicating issue reports, and asking engineers for status updates in Slack.

Dust bridges this gap not through rigid point-to-point synchronization rules, but through contextual AI orchestration. By connecting both Zendesk and GitHub to Dust, teams create custom AI agents that understand both customer conversation history and technical codebases. These agents can automatically triage incoming support tickets, detect existing GitHub issues, draft internal troubleshooting notes, and file standardized engineering bugs without human context switching.

Here is how to design, configure, and deploy an end-to-end triage integration using Dust, Zendesk, and GitHub.

Architectural Overview: How Dust Unifies Support and Engineering

To understand how Dust connects Zendesk and GitHub, it is helpful to distinguish between the three functional layers Dust uses to interact with external platforms:

This separation ensures that Dust is not merely syncing raw fields back and forth. Instead, it reads technical context from GitHub, evaluates customer symptoms in Zendesk, and takes reasoned actions across both tools.

Step 1: Establish Workspace Data Connections

Before agents can perform cross-platform triage, Dust needs read access to your historical support knowledge and engineering issue logs.

1. Connecting Zendesk to Dust

1. In your Dust workspace, navigate to Spaces > Connections and click Zendesk.
2. Provide your Zendesk instance subdomain (https://<subdomain>.zendesk.com).
3. Complete the OAuth authorization as a Zendesk administrator with access to both Support and Guide.
4. Open the connection settings under Manage to configure data filters:
   • Ticket Scope & Retention: Dust defaults to syncing solved and closed tickets updated within the last 180 days, with a rolling retention cleanup of 120 days. You can optionally toggle Sync Unresolved Tickets if agents must reference active work-in-progress incidents.
   • Customer Privacy (PII Protection): Enable Hide Customer Information to strip requester names and email addresses from synced metadata while preserving the ticket technical content.
   • Tag and Field Filtering: Apply Included Tags or Excluded Tags to restrict sync to technical ticket queues (e.g., tags like tier-2 or bug). If your team tracks incident IDs or impacted microservices in custom Zendesk fields, add the numeric Field ID under Custom Field Tags so Dust maps them to searchable labels in the format fieldName:value.
   • Rate Limiting: If your Zendesk subscription has strict API caps, configure a transaction-per-second (TPS) ceiling.

2. Connecting GitHub to Dust

1. In Dust, go to Spaces > Connections > GitHub and click Connect.
2. Complete GitHub authentication as an organization administrator.
3. Select the exact repositories Dust should access (e.g., your core application backend, frontend client, and public SDKs).
4. Optionally toggle Code synchronization if you want your triage agents to inspect default-branch source files in addition to issue threads and pull requests.
5. Save the configuration. Dust will index issues, discussions, and top-level pull request comments in real time, while full codebase indexing refreshes on an 8-hour cycle.

Step 2: Enable Zendesk and GitHub Tools

Data connections give Dust read-only knowledge. To allow agents to search active tickets, write private triage summaries, and file GitHub issues, you must activate the corresponding Tools:

1. Navigate to Spaces > Tools and select Add Tools.
2. Select Zendesk and complete workspace-level OAuth. This unlocks:
   • Search Tickets: Executes query syntax (e.g., status:open priority:urgent tags:bug).
   • Get Ticket: Retrieves complete comment history, metadata, and SLA response metrics.
   • Draft Reply: Generates private internal comments on a ticket without publishing them to the customer.
3. Select GitHub and complete workspace authorization. Choose whether actions should run under:
   • Workspace Credentials: All GitHub actions, issues, and comments are posted by the @dust-agent GitHub bot. This is the recommended choice for automated triage pipelines.
   • Personal Credentials: Actions are executed on behalf of the individual user invoking the agent, respecting that specific user's repository rights.

Step 3: Build the Triage and Escalation Agent

With connections and tools active, open Dust's Agent Builder to create an agent dedicated to support-to-engineering triage (e.g., named @tier2-triage).

Agent Configuration Checklist

• Model: Select an advanced reasoning model capable of analyzing technical logs and code snippets (e.g., Claude 3.5 Sonnet, GPT-4o, or Gemini 1.5 Pro).
• Knowledge Sources: Grant access to your Zendesk Connection (Help Center and past tickets) and GitHub Connection (repository issues and code).
• Tools: Enable Zendesk (Search Tickets, Get Ticket, Draft Reply) and GitHub (Create Issue, Search Issues).
• Instructions: Define explicit rules for how the agent diagnoses problems, links existing defects, and escalates unresolved bugs.

You are the Tier-2 Support and Issue Triage Copilot. Your job is to analyze incoming Zendesk customer support tickets, cross-reference them with our GitHub repositories, and provide actionable technical triage.

When analyzing a Zendesk ticket:
1. Search our GitHub repositories (default: dust-tt/app-backend) to check whether this error, stack trace, or unexpected behavior matches an existing open or recently closed issue.
2. Search historical Zendesk tickets to identify whether this issue is a recurring customer-facing symptom with a verified workaround.
3. If an existing GitHub issue matches:
   - Use Zendesk "Draft Reply" to post a private internal note.
   - Include: (a) GitHub Issue URL and current issue state, (b) Root cause summary, and (c) Recommended technical explanation or workaround to share with the user.
4. If this is a novel bug and confirmed reproducible:
   - Create a new GitHub issue in the target repository using this structure:
     - Title: [Triage] Brief description of failure
     - Body: Impacted Zendesk ticket ID, customer-reported symptoms, environment/browser details, and full error logs.
   - Post a private internal note in Zendesk referencing the newly created GitHub issue ID.
5. Never post public comments to the end customer. All outputs must be drafted as private internal notes or formatted responses in the Dust sidebar.

Step 4: Deploy the Interface and Event Triggers

Teams can operate this integration in two complementary modes: on-demand within the Zendesk workspace, and autonomously via event webhooks.

Mode A: Frontline Copilot in the Zendesk Sidebar

1. Install the official Dust application from the Zendesk Marketplace.
2. In the Zendesk Admin Center, open the Dust app settings.
3. Paste your Dust Workspace ID and an API Key generated by a Dust workspace administrator.
4. In the Default Agent IDs setting, add the ID of your @tier2-triage agent. This keeps your support agents focused on the designated triage assistant directly beside the customer ticket timeline.
5. Configure role or group restrictions to ensure only authorized support representatives and technical leads have access.
6. Support agents can now open any customer inquiry, click the Dust sidebar, and immediately ask: "Has engineering addressed this Stripe checkout timeout before?" The agent searches GitHub, checks relevant commit messages or open issues, and produces an instant diagnostic summary.

Mode B: Fully Automated Webhook Triage

To triage high-priority tickets before a support agent even opens them:

1. In Dust, go to Spaces > Triggers > Add source.
2. Select Zendesk as a built-in webhook provider and connect your account.
3. In the Agent Builder for @tier2-triage, go to Triggers > Add triggers and pick the Zendesk webhook source.
4. Configure trigger rules:
   • Listen for: Ticket created or updated.
   • Run only when: Specify plain-language conditions such as "priority is high or urgent and tags include engineering-review".
   • Include webhook payload: Toggle this on so the ticket ID, subject, and comment body pass directly to the agent.
   • Rate limit: Set a safety ceiling (e.g., maximum runs per 24 hours) to protect API quotas during unexpected ticket volume spikes.
5. Save the agent. When an escalation occurs, Dust triggers in the background, pulls repository context from GitHub, and automatically posts a private internal draft to the Zendesk ticket.

Operational Comparison: Native Dust vs. Basic Sync Plugins

Governance, Pricing, and Best Practices

1. Keep Human-in-the-Loop on External Communication: Always configure Dust agents to use Draft Reply rather than sending public messages. A human support agent should verify that engineering issue details and estimated fixes are appropriate to communicate to external customers.
2. Prevent API Quota Exhaustion: Monitor Zendesk API usage. When setting up webhooks, use narrow trigger filters (such as specific priority levels or support queues) rather than firing an LLM evaluation on every single password reset ticket.
3. Plan for Team Tiers and Credit Consumption: Dust Business plans operate on self-serve tiers, including Pro at €24/seat/month (8,000 credits/seat/month) and Max at €120/seat/month (40,000 credits/seat/month), billed yearly. Enterprise plans provide workspace-pooled credits, SCIM provisioning, custom audit logs, and dedicated SLA support. Work with your team leads to allocate credits appropriately across automated webhook triggers and interactive sidebar inquiries.

By uniting Zendesk's customer ticket flow with GitHub's developer records through Dust, technical support organizations eliminate duplicate tickets, accelerate incident escalation, and maintain alignment between engineering and customer success teams.

References

• Zendesk Tools - Dust Docs
• GitHub Tools - Dust Docs
• Zendesk Connection - Dust Docs
• GitHub Connection - Dust Docs
• Dust in Zendesk - Dust Docs
• Webhooks - Dust Docs
• Customer Support - Dust Docs
• Dust Pricing: Business and Enterprise Plans for AI Agents