Goal
Give an AI agent repository-aware support context so it can answer questions with fresher product and issue knowledge.Before you start
- You should have a published or near-ready support agent.
- Decide which repositories are safe and useful for support context.
Step 1: connect the GitHub app
In the GitHub integration area:- connect the GitHub app installation
- choose the repositories your org wants Sippet to access
- confirm the installation shows healthy before binding repos
Step 2: create repository bindings
For each useful repository:- add the repository binding
- confirm the default branch and repo status look correct
- trigger or review sync status if the repo is newly connected
Step 3: assign the repository to an agent
Use agent bindings to decide which support agent should use which repo context. This is useful when:- one agent handles one product area
- different teams own different repositories
Step 4: validate the support behavior
Test questions like:- “Is this issue already known?”
- “Did this feature change recently?”
- “What should support tell the customer while engineering is still working?”
Example workflow
- a support agent is linked to your main product repository
- the agent answers a user question using repository knowledge and issue context
- if the issue is unresolved, the support workflow can escalate toward issue reporting or human triage
What good looks like
- repository sync stays healthy
- support answers reflect the correct product area
- agents use GitHub context without sounding like raw engineering output
Troubleshooting
- If answers feel stale, review repository sync status.
- If the wrong product context appears, narrow the agent-to-repository bindings.