Turn a support report into an engineering issue
Combine a user-provided failure report with relevant repository documentation and existing issue context. Prepare reproduction steps and unknowns before creating the issue.
GitHub integrationAI agents for GitHub issues and repository context
Find repository context, prepare actionable issues, and open pull requests between existing branches with ExemplaryAI agents.
Connected actions
What your agent can do with GitHub
The ExemplaryAI GitHub integration gives agents actions for finding repositories, reading files, searching code, listing issues, and creating issues or pull requests. Use that context to turn an incomplete request into a clearer engineering handoff. Keep the repository, source material, and proposed change explicit so the team can review what the agent found and what it should create.
Find repositories
List repositories available to the connected account and identify the intended owner and project.
Read repository details
Retrieve a repository's details to establish the project context before using other actions.
Inspect existing issues
List issues in the selected repository to inform a handoff. Review returned results before calling a request new or a duplicate.
Search repository code
Search accessible code for a relevant term or symbol, then inspect the appropriate source file.
Read a source file
Retrieve file contents to answer a focused question about the code or documentation, without claiming to have executed it.
Create an issue
Create a reviewed issue in the chosen repository with a clear problem statement and supporting details.
Open a pull request
Create a pull request between existing branches. The action does not create the underlying code changes or branches.
Where it helps
Put GitHub context to work
Answer a focused repository question
Search for a configuration name, read the matching file, and explain the relevant behavior. Keep conclusions tied to the source that was actually retrieved.
Prepare an existing branch for review
Use a developer-provided change summary to prepare a pull request between confirmed existing branches. The developer remains responsible for the changes and testing evidence.
Worked example
A clearer issue from an incomplete report
Illustrative example, not a recorded product run. A support report says a CSV export has missing column headers, but provides no reproduction steps.
Confirm the repository and inspect the returned issue list for relevant reports.
Search for export documentation and read the appropriate file.
Prepare the issue text with the observation, relevant source, and questions for the reporter.
Illustrative output
- Proposed title
- Investigate missing column headers in CSV export
- Known evidence
- A user supplied the symptom; the example does not include the affected file or application version.
- Needed next
- Request the version, export settings, and a suitable sample before claiming a reproducible defect.
Check for duplicates, sensitive content, and the correct repository. Issue creation is a separate requested action after this review; preparing the text does not establish an enforced approval workflow.
Connection setup
Connect GitHub to your agent
Start with a defined task, the account that should perform it, and the smallest set of actions it needs.
Select the repository task
Choose a focused issue or repository question and record the intended owner and repository name.
Review OAuth authorization
Configure a GitHub OAuth app with the deployment's callback URL and client credentials. Review its requested scopes and organization policies with the account owner before authorization.
Understand the access breadth
The default repo scope is broad. Confirm the connected account's access; choosing a repository in a prompt is not a token-level restriction.
Test reading before creation
Check retrieved context first. For a write test, review the issue text or existing head and base branches before requesting creation.
Access & deployment
Set the connection boundary
- Authorization
- GitHub OAuth scopes and the connected user's permissions govern access; this is not a GitHub App installation with per-repository grants.
- Repository scope
- The repo OAuth scope can grant broad read and write access to public and private repositories the user can access.
- Requested actions
- A task's target repository should be explicit. Agent instructions alone do not narrow the underlying OAuth grant.
Context is not execution evidence
Reading code does not prove that tests pass or a reported problem is reproducible. Attribute conclusions to the files and reports inspected.
Existing branches are required
These actions do not establish code editing, branch creation, pull-request review, or merging. A pull request needs existing branches and prepared changes.
GitHub.com connectivity
The documented connection needs permitted access to GitHub's cloud APIs. Private agent hosting does not provide strict air-gap operation or establish GitHub Enterprise Server support.
GitHub integration questions
Can the GitHub AI agent fix code automatically?
These integration actions provide repository reading and selected creation actions. They do not establish code editing or an autonomous fix-and-test workflow.
Can it create a GitHub issue?
Yes, when the connected account has permission. Provide the intended repository and review the proposed issue content before requesting creation.
Can it create a pull request?
Yes, between existing branches. Confirm the head and base branches and provide accurate testing information; creating the pull request does not generate its changes.
Is access limited to the repository in my request?
The request identifies where the agent should work. The OAuth grant can be broader, so review the account's permissions and scopes separately.
Keep exploring
Plan the wider workflow
Provider references
Use these provider documents when reviewing account permissions and API behavior.
Reviewed
See Exemplary AI in action
Book a demo and we'll show you purpose-built agents grounded in your own knowledge — deployed on your infrastructure.