GitHub Pull Requests and Issues in VS Code

The form-github-search-query skill converts natural language queries into properly formatted GitHub search syntax for issues and pull requests. It allows users to leverage GitHub's advanced search capabilities (like filtering by assignee, label, or state) without needing to manually construct complex query strings.

To use this skill, provide a natural language query and specify the search type (issue or pr).

If a comment contains a suggestion block, you can apply it directly to the file:

1. Focus on the existing comment.
2. Use the Apply Suggestion button (accessible via Tab) to apply the change.

Note: Applying a suggestion only modifies the local file. To update the Pull Request on GitHub, you must manually commit the file change and push it to the remote branch.

To begin using the extension to review and manage GitHub pull requests and issues in VS Code, follow these steps:

1. Install the extension: Search for it within VS Code or download it from the VS Code Marketplace.
2. Open a repository: Open a GitHub repository in VS Code.
3. Access the viewlet: A new viewlet will appear on the Activity Bar showing pull requests and issues.
4. Authenticate: Click the sign-in button on the viewlet to connect to GitHub or GitHub Enterprise.
5. Configure remotes (if necessary): By default, the extension looks for PRs on origin and upstream. If your repository uses different remotes, configure the githubPullRequests.remotes setting.

From the Issues view, you can select Start Working to automate the setup for an issue. This action creates a new branch and populates the commit message.

Customizing Branch Creation

• Disable branch creation: Set "githubIssues.useBranchForIssues": "off".
• Prompt for branch name: Set "githubIssues.useBranchForIssues": "prompt".
• Custom branch naming: Use "githubIssues.issueBranchTitle": "${user}/issue${issueNumber}" to define a naming pattern.

When creating an issue, you can use a text editor to modify properties by clicking the pencil icon. The editor uses specific conventions to set:

• Issue title
• Assignees
• Labels
• Body

To edit other properties, you must open the issue directly on GitHub.

The suggest-fix-issue skill is designed to propose potential solutions for a given issue by analyzing its details. It processes the issue's description, comments, and any available relevant context to generate a targeted suggestion for resolution.

To use this skill, provide the issue details (description, comments, and context). It is particularly useful for troubleshooting guidance or for developers seeking a starting point for addressing a problem.

When using the suggest-fix-issue skill or manually proposing solutions, follow these guidelines to ensure high-quality suggestions:

• Understand the context: Thoroughly read the issue's description and all comments to ensure a full understanding of the problem.
• Provide actionable code: Where possible, output code blocks and reference real files within the workspace that contain the proposed fix.
• Address the root cause: If the issue is a bug, identify the underlying root cause and suggest a fix that addresses it directly rather than just treating the symptoms.

You can create GitHub issues directly from TODO comments in your code using CodeLens (clickable actions above the line) or Code actions (via the lightbulb menu).

Available actions:

• Create Issue from Comment: Creates an issue with your text selection as a permalink in the body and inserts the issue number after the TODO.
• Delegate to coding agent: Starts a Copilot coding agent session to work on the task (when available).

By default, the triggers are TODO, todo, BUG, FIXME, ISSUE, and HACK. You can customize these using the githubIssues.createIssueTriggers setting. You can also use the Create Issue from Selection command to create an issue from any selected text.

When converting natural language to search syntax, follow these rules for label selection:

• Relevance: Choose labels based on the user's intent, not just matching words. If unsure, it is better to have no labels than irrelevant ones.
• Specificity: Avoid picking multiple highly specific labels; labels are and-ed together. Aim for one relevant label.
• Exclusion: Do not include labels that the user has explicitly excluded.
• Context: Words in inline codeblocks in a query are likely to refer to labels. Match these against the actual labels in the repository.

The extension provides autocomplete suggestions for GitHub entities:

Issues

Trigger by typing # or running the Trigger Suggest command.

• Behavior: In most files, it inserts the issue number. In Markdown, it inserts a markdown link (e.g., [#1234](url)).
• Configuration:
  • Change SCM commit message format with githubIssues.issueCompletionFormatScm (e.g., use "${issueNumberLabel}").
  • Ignore the # trigger with githubIssues.ignoreCompletionTrigger.
  • Ignore specific milestones with githubIssues.ignoreMilestones.

Users

Trigger by typing @ or running the Trigger Suggest command. Selecting a user inserts their username (e.g., @alexr00).

The show-github-search-result skill is used to transform raw GitHub search query results (issues or pull requests) into a human-friendly markdown table. This skill is designed to summarize key details like title, author, labels, and state, allowing users to review search results at a glance.

Note: This skill is not user-invokable (user-invokable: false) and is intended to be used by the system to process raw search data.