qa

You can convert conversational bug reports into durable GitHub issues by providing the expected behavior, actual behavior, and reproduction details. The agent asks up to three clarifying questions, explores the codebase for domain language, and creates issues with mandatory reproduction steps written from the user's perspective.

Codebase exploration helps scope multi-symptom QA reports by running a background subagent to learn the feature's intent and ubiquitous domain language. This ensures the resulting GitHub issues use accurate terminology without requiring the reporter to manually map symptoms to code paths.

Yes, you can create multiple independently fixable GitHub issues from a single bug report. When a multi-symptom QA report requires scoping, the agent separates the defects into distinct issues, each with clear expectations and mandatory reproduction steps created via the gh issue create command.

No, the generated GitHub issues strictly avoid code and file-path references. The agent writes issues from the user's perspective using domain terminology learned during background codebase exploration, ensuring the report remains durable and focused on expected versus actual behavior.

You need to provide a description of the bug, what you expected to happen, what actually happened, and how you tried to reproduce it. The agent handles the rest, asking light clarification questions and creating the issue using the gh issue create command.

Using an agent to scope bug reports ensures issues include mandatory reproduction steps and domain terminology extracted from codebase exploration. This approach prevents inconsistent behaviors from being poorly documented and returns formatted GitHub issue URLs directly to the user.