DEV Community: Öcal Esmer

What I Want to See Before Merging an AI-Generated PR

Öcal Esmer — Sun, 01 Nov 2026 10:46:14 +0000

When a coding agent finishes, “done” leaves me with questions.

What changed? Did the project build? Which tests ran? Does the result address the original acceptance criteria?

Those questions shaped how I built Coreaven, a workflow tool for .NET projects.

1. A diff tied to the task

The first thing I want to inspect is the actual change. A summary helps me navigate, but I still need to review the code.

Coreaven connects the task, reviewed prompt, execution, and resulting GitHub PR so I can follow the work from request to result.

2. Explicit validation results

A successful agent process does not tell me whether the project builds or its tests pass.

I want to see the restore/build/test results, including failures and skipped steps. A skipped check should remain visible as skipped.

3. Evidence for individual acceptance criteria

A passing build does not prove that every requirement was implemented.

Coreaven shows acceptance-criteria evidence as Verified, Partial, or No Evidence. These states help identify what has supporting validation and what still needs attention.

They support review; they do not replace it.

4. A clear review decision

Coding runs through a Windows Runner in the user’s environment, using their own coding-agent account.

After execution, the developer can inspect the changes and validation results, review the GitHub PR, and decide whether to merge.

Try it on a real task

I’m looking for .NET developers willing to try one small task on an existing repository. I can help with setup and the first run.

https://coreaven.com

What evidence do you look for before merging an AI-generated PR?

Disclosure: I’m the developer behind Coreaven. This post was drafted with AI assistance.

From a Jira Issue to a GitHub PR: The Coreaven Workflow for .NET

Öcal Esmer — Thu, 22 Oct 2026 13:17:10 +0000

Generating code is one part of an AI coding task. Preparing context, reviewing instructions, checking the result, and updating the original issue are part of the work too.

I’m building Coreaven to keep those steps connected for .NET projects.

The workflow

Create a task or import an issue/card from Jira or Trello.
Generate a prompt using repository context from Codebase Brain, Coreaven’s Roslyn-based .NET code index.
Review and edit the prompt and acceptance criteria before starting execution.
Run your coding agent through a Windows Runner in your environment, using your own agent account. Coding happens in an isolated Git worktree.
Follow live execution progress.
Inspect the diff, restore/build/test results, and evidence associated with the acceptance criteria.
Review the resulting GitHub PR before deciding whether to merge.
Send the result summary and PR link back to the original Jira issue or Trello card.

Importing an issue does not automatically start coding. You review the prompt and choose when to run the agent.

Try a small task

Setup currently requires connecting GitHub, installing the Windows Runner, and authenticating your coding agent. I’m happy to help developers through setup and a first task.

Have a small task in an existing .NET repository? Let’s try the workflow together.

https://coreaven.com

What information do you need alongside an AI-generated PR to make reviewing it easier?

Disclosure: I’m the developer behind Coreaven. This article was drafted with AI assistance.

Why I Built Coreaven to Connect .NET Coding Tasks with GitHub PRs

Öcal Esmer — Wed, 21 Oct 2026 08:23:52 +0000

I’m Öcal, the developer behind Coreaven. I built it to keep the steps around an AI coding task connected: requirements, repository context, prompt preparation, execution, validation results, and pull request review.

This post introduces my own product and the engineering choices behind its initial .NET workflow.

Start with a task you can review

The workflow starts with a task and optional acceptance criteria. Coreaven prepares a repository-aware prompt that you can inspect and edit before running your coding agent.

For an existing .NET codebase, the relevant context includes more than a file name: symbols, their relationships, repository instructions, and the exact revision being changed all matter. Coreaven uses a Roslyn-based code index to help select that context.

Keep execution separate from the working checkout

The coding agent runs through a Windows Runner in your environment, using your own coding-agent account. Each coding job uses an isolated Git worktree.

That gives the task a specific base commit and a separate workspace for its changes. You can follow execution activity and inspect the resulting files without treating “the agent finished” as proof that the task succeeded.

Make validation visible

For the .NET workflow, validation follows restore → build → test. The results remain associated with the execution so you can review them alongside the changes.

A passing build and test run is useful evidence, but it does not establish that every requirement is met. Acceptance criteria and the actual diff still need review. Coreaven keeps validation, review, and GitHub delivery visible as separate stages.

Review the result before merging

The complete path is:

Task → repository-aware prompt → coding execution → validation → GitHub PR → human review

You review the prompt before execution and the changes before merging. The goal is to make it easier to trace a pull request back to the task that produced it and the validation that ran.

What I’m working on next

The current onboarding includes connecting GitHub, installing the Runner, and authenticating the coding agent. That setup is a real part of the experience, and reducing its friction is one of my priorities.

I’m looking for .NET developers and small teams willing to try one small task on an existing repository. I’m happy to help with setup and the first run.

You can explore Coreaven at https://coreaven.com.

What information do you want alongside an AI-generated PR to make the review useful? I’d especially like feedback on repository context, validation results, and the handoff to human review.