Code Suggestions

GitSniff generates code suggestions using AI models that understand your codebase, coding style, and the specific issue being addressed.

How suggestions are generated

1. Issue identification — The AI model identifies a specific issue in the code
2. Context gathering — GitSniff considers the surrounding code, file structure, and repository conventions
3. Fix generation — The model generates a corrected version that addresses the issue
4. Validation — The suggestion is checked for syntax correctness before posting

Custom instructions

Auto-fix respects your repository's custom instructions. If you have a .opencode/instructions.md file in your repository, GitSniff will follow those guidelines when generating suggestions.

This means suggestions will:

• Follow your project's coding standards
• Use your preferred patterns and libraries
• Respect language-specific conventions you've defined

See Custom Instructions for setup details.

Applying suggestions

To apply a suggestion:

1. Review the suggestion in the PR comment
2. Click Apply suggestion in the GitHub UI
3. The change is applied as a new commit

You can also batch-apply multiple suggestions at once using GitHub's batch suggestion feature.