Standard Mode

Standard mode is CommitStudio's primary analysis tool, providing in-depth AI-powered code review for your Git repositories. This mode analyzes code changes, generates insightful feedback, and posts comments directly to GitHub.

How Standard Mode Works

CommitStudio's standard mode follows a sophisticated multi-step process:

1. Repository Detection

   • Detects the Git repository from the current directory or specified path
   • Identifies the GitHub remote and repository owner/name
   • Verifies access permissions with your GitHub token

2. Commit Selection

   • Retrieves commits based on your filtering options (count, branch, date, author)
   • Identifies which commits have already been processed (using cache)
   • Prepares unprocessed commits for analysis

3. Diff Generation

   • Uses Git to extract the diff for each commit
   • Formats diffs to optimize for AI analysis
   • Ensures context is preserved for accurate review

4. AI Analysis

   • Sends the formatted diff to OpenAI's API
   • Uses a specialized prompt that instructs the AI to:
     • Identify bugs, security issues, and code smells
     • Suggest improvements and best practices
     • Consider language-specific conventions
     • Provide constructive, actionable feedback
   • Processes the response into structured comment format

5. GitHub Integration

   • Connects to GitHub using your personal access token
   • Posts AI-generated comments on the specific commits
   • Formats comments with Markdown for readability
   • Ensures comments are properly attributed to CommitStudio

6. Results Caching

   • Stores processed commits in a local cache
   • Avoids reanalyzing the same commits in future runs
   • Optimizes performance and reduces API costs

7. Feedback Presentation

   • Displays a summary of the analysis in the terminal
   • Shows progress indicators during analysis
   • Provides links to the GitHub comments

When to Use Standard Mode

Standard mode is ideal for:

• Daily Development: Regular code review during development
• Pre-PR Review: Catching issues before creating pull requests
• Team Collaboration: Sharing AI insights with your team
• Quality Assurance: Ensuring code meets quality standards
• Knowledge Transfer: Getting explanations of complex code

Command Syntax and Options

Basic Usage

# Use default settings
commitstudio

# Specify repository path
commitstudio --path /path/to/repo
Copy code

Commit Selection

# Analyze specific number of commits
commitstudio --commits 5

# Analyze commits on a specific branch
commitstudio --branch feature/new-feature

# Analyze commits since a date
commitstudio --since "2023-01-01"

# Analyze commits from a specific author
commitstudio --author "user@example.com"
Copy code

Execution Options

# Analyze without posting to GitHub (preview mode)
commitstudio --dry-run

# Ignore cache and reanalyze all commits
commitstudio --no-cache

# Show detailed logs during analysis
commitstudio --verbose
Copy code

Advanced Configuration

# Reset all saved settings and credentials
commitstudio --reset

# Combine multiple options
commitstudio --commits 10 --branch develop --dry-run --verbose
Copy code

How AI Analysis Works

CommitStudio leverages advanced AI models to analyze your code:

The Analysis Pipeline

1. Preprocessing: The diff is prepared with appropriate context
2. Tokenization: The code is converted to tokens for the AI model
3. AI Processing: The model analyzes the tokens using its trained knowledge
4. Response Generation: The AI generates structured feedback
5. Postprocessing: The response is formatted into actionable comments

What the AI Evaluates

• Logical Errors: Bugs, edge cases, and potential failures
• Security Issues: Vulnerabilities, injection risks, and unsafe practices
• Performance Concerns: Inefficient algorithms and resource usage
• Code Quality: Maintainability, readability, and best practices
• Architecture: Design patterns and structural improvements
• Documentation: Missing or unclear documentation

AI Model Selection

The quality of analysis depends on the AI model used:

• Default: gpt-4.1-mini (balanced performance and cost)
• For Critical Code: Consider upgrading to gpt-4o or gpt-4.1
• For Large Repositories: Models with higher token limits help analyze more code

To change the model:

commitstudio config --model gpt-4o
Copy code

GitHub Integration Details

CommitStudio posts comments directly to your GitHub repository:

Comment Placement

• Commit Comments: Posted on the specific commit that was analyzed
• Line-Specific: Comments tied to the relevant lines of code
• Grouped: Related issues are combined when appropriate

Comment Formatting

• Markdown Support: Rich formatting with code blocks and lists
• Issue Categorization: Clear labeling of issue types
• Code Context: Relevant code snippets included
• Suggested Solutions: Actionable recommendations

Permissions Required

• Your GitHub token must have the repo scope
• You must have write access to the repository
• CommitStudio respects repository permissions

Real-World Example

Here's a typical terminal output when running standard mode:

✓ Repository detected: username/project-name
✓ GitHub connection established
✓ Found 7 commits to analyze (3 already processed)
✓ Analyzing diffs with AI...
  ↪ Analyzing commit 1/4: Add user authentication (65d7a9f)
  ↪ Analyzing commit 2/4: Fix pagination bug (18e2b3c)
  ↪ Analyzing commit 3/4: Implement search feature (24f1d8e)
  ↪ Analyzing commit 4/4: Update dependencies (93c7b5a)
✓ Posting comments to GitHub...
  ↪ Posted 3 comments on commit 65d7a9f
  ↪ Posted 1 comment on commit 18e2b3c
  ↪ Posted 2 comments on commit 24f1d8e
  ↪ Posted 0 comments on commit 93c7b5a
✓ CommitStudio completed successfully!

📊 Summary: 4 commits analyzed, 6 comments posted
Copy code

And on GitHub, comments appear like:

## 🤖 CommitStudio AI Review

### Potential Security Issue: SQL Injection Risk

The user input from `searchQuery` is directly concatenated into the SQL query,
creating a potential SQL injection vulnerability:

```js
const query = `SELECT * FROM users WHERE name LIKE '%${searchQuery}%'`;
Copy code

Recommendation: Use parameterized queries to prevent SQL injection:

const query = `SELECT * FROM users WHERE name LIKE ?`;
const params = [`%${searchQuery}%`];
Copy code

This ensures user input is properly escaped and keeps your database secure.

## Optimizing Your Experience

To get the most out of standard mode:

1. **Make Focused Commits**: Smaller, focused commits get better analysis
2. **Use Descriptive Messages**: Helps the AI understand context
3. **Review Incrementally**: Address issues as they're identified
4. **Configure AI Model**: Choose the right model for your needs
5. **Use Dry Run**: Preview analysis before posting to GitHub

## Related Topics

- [Configuration Options](/docs/configuration/options) - Customize AI models and behavior
- [YOLO Mode](/docs/usage/yolo-mode) - Learn about CommitStudio's commit message rewriting feature
- [CI/CD Integration](/docs/advanced/ci-cd-integration) - Automate code reviews in your pipeline 
Copy code