Preserve customer language
Selected source messages appear as quoted evidence, while the edited Zens summary explains the problem in product language. Engineers can see what the customer actually said and what the support team concluded.
Zens AI GitHub App
Customer evidence that engineering can act on
The Zens AI GitHub customer support integration turns a verified support conversation into a structured GitHub issue. Product teams receive the customer’s words, affected path, account context, and a direct link back to Zens AI instead of a vague bug summary copied between tools.
Issues read and write, plus read-only repository metadata
acme/dashboardIssues / #284
Checkout loses the selected annual plan after sign-in
Opened by Zens AI · customer report · bug
Customer-reported problem
The pricing page shows the annual discount, but checkout returns to monthly billing after account creation.
Original evidence
“I chose annual, signed in, and the checkout now says monthly.”
Reported path: /pricingCustomer plan: Trial
More precise than a generic automation
A generic webhook can copy a chat message into GitHub, but it cannot decide whether the message is the actual product problem, an agent reply, or unrelated account detail. Zens AI creates a reviewed issue record first. The GitHub handoff uses that saved record, so the title, summary, evidence, type, and product path have already been checked by the team that spoke with the customer.
The result is also traceable in both directions. The GitHub issue links to Zens for authorized conversation context, while Zens stores the GitHub repository, issue number, URL, and current state. Teams can answer a practical question later: which customer report became which engineering task, and did GitHub show that work as completed?
A shorter path to the fix
Keep the report attached to its source
A typical customer support handoff loses detail at every step. Zens AI keeps the original evidence beside the GitHub issue, so engineering can judge impact and reproduce the behavior without reopening the entire customer interview.
- 01
Capture a useful conversation
Zens AI records the support thread, signed customer identity, current page, and product events. A teammate chooses which source messages are relevant instead of exporting the full conversation.
- 02
Shape a product issue
The team reviews the title, summary, issue type, and internal note. The original AI draft remains available for audit, while the saved version becomes the engineering brief.
- 03
Choose an approved repository
The GitHub App lists only repositories granted during installation. Private repositories are preferred. A public repository requires an explicit warning and confirmation before customer evidence can be posted.
- 04
Create the GitHub issue
Zens AI sends the saved problem statement, selected evidence, reported path, and a reproduction starting point. Customer email and external user IDs are deliberately omitted from the GitHub issue.
- 05
Track the result
Signed GitHub webhooks keep the linked issue state current when engineers edit, close, or reopen it. The Zens record continues to connect the customer report with the work that followed.
Built for evidence, not ticket volume
A GitHub customer support integration for product teams
Carry product context
The GitHub issue includes the reported route, conversation status, issue type, and—only for private repositories—limited account and plan context. The source link opens the full Zens record for authorized teammates.
Respect repository visibility
The GitHub App never assumes that customer text is safe for a public repository. Public export stays blocked until a teammate reviews the evidence and explicitly approves publishing it.
Minimal GitHub permissions
The Zens AI GitHub App requests Issues read and write so it can create an issue and follow its open or closed state. GitHub supplies read-only Metadata automatically so Zens can show the repository name and visibility. The integration does not request source code, pull request, Actions, deployment, administration, organization member, email, or secret access.
Installation tokens are generated only on the server, expire after one hour, and are never stored in the browser. Every webhook is verified with HMAC-SHA256 and recorded by delivery ID so GitHub retries cannot duplicate processing. Uninstalling the GitHub App disables repository access and stops synchronization.
Useful across the product loop
Support teams
Escalate a bug without rewriting the customer story in another form.
Product managers
Group support evidence into a reviewed issue before engineering work begins.
Engineers
Start with the affected path, real customer wording, and a bounded reproduction brief.
Founders
Connect repeated customer friction to the GitHub work that resolves it.
GitHub integration questions
Can Zens AI read our source code?
No. The GitHub App does not request Contents permission. It receives the repository metadata needed to identify an approved destination and the Issues permission needed to create and synchronize the linked issue.
What appears in a GitHub issue?
The issue contains the saved Zens summary, selected source messages, issue type, reported path, a reproduction starting point, and a secure link back to Zens AI. Email addresses and external user identifiers are omitted.
Does the integration work with private repositories?
Yes. Install the GitHub App for the organization or personal account, then grant access to all repositories or selected repositories. Zens lists only the repositories included in that installation.
How do I remove the integration?
Uninstall Zens AI from the GitHub account’s installed-app settings. GitHub sends a signed uninstall event, Zens disables the repository connection, and no new installation tokens can be created.