Creating & Managing Issues

Step-by-step guide to creating issues manually, setting priorities, editing issue details, and managing issues in bulk.

5 min read
issuescreatemanage

While issue sessions can discover problems automatically, you can also create and manage issues manually. This is useful when you already know what needs to be fixed, want to add context to AI-discovered issues, or need to organize work items that don't require codebase scanning.

Creating an Issue Manually

To create an issue from the Issues page, click the New Issue button. The manual issue dialog opens with the following fields:

  1. Title (required) - A short, descriptive summary of the problem or task. Keep it concise but specific enough to identify the issue at a glance.
  2. Description (required) - A detailed explanation of the issue. Include context about where the problem occurs, what the expected behavior should be, and any relevant error messages or symptoms.
  3. Priority (required) - The importance level of the issue. Defaults to medium if not specified.
  4. Suggested Prompt(optional) - Specific instructions for an AI agent to resolve the issue. If provided, this prompt is used instead of the description when launching a workflow run. Write it as a direct instruction (e.g., "Fix the memory leak in src/ws/handler.ts by cleaning up event listeners on disconnect").

Writing Good Suggested Prompts

A well-crafted suggested prompt dramatically improves AI resolution success. Include the specific file paths, function names, and the desired fix approach. The more precise the instructions, the better the result.

Priority Levels

Every issue has a priority level that helps you organize and triage work:

PriorityWhen to Use
criticalProduction-breaking bugs, security vulnerabilities, or data loss scenarios that require immediate attention.
highSignificant bugs or issues that affect core functionality but have workarounds. Should be addressed in the current sprint.
mediumNon-critical improvements, minor bugs, or tasks that should be done but aren't urgent. This is the default priority.
lowNice-to-have improvements, code cleanup, or minor cosmetic issues that can wait.

Editing Issues

You can update any issue's fields after creation. From the issue detail view, edit:

  • Title - Refine the summary as you learn more about the problem
  • Description - Add additional context or investigation notes
  • Priority - Adjust based on impact or urgency changes
  • Suggested Prompt - Update fix instructions based on new findings

Running Issues

Issues with status running should not be edited, as changes will not affect the active workflow run. Wait for the run to complete or fail before making changes and retrying.

Deleting Issues

Issues support soft deletion - deleted issues are moved to trash and can be recovered if needed. To delete issues:

  • Single delete - Click the delete action on an individual issue card or from the issue detail view.
  • Bulk delete - Select multiple issues using the checkboxes on the issues list, then click the bulk delete action. This is useful for cleaning up large batches of irrelevant AI-discovered issues.

Discovering Issues with AI

For automatic issue discovery, CodeCourier provides issue sessions - AI-powered scanning runs that analyze your codebase and extract issues automatically. An issue session provisions a cloud sandbox, runs an AI agent with your discovery prompt, and produces a structured list of issues complete with titles, descriptions, priorities, and suggested fix prompts.

Next Steps