AI Commands¶

AI-powered CLI commands for development.

Overview¶

Aksara's AI commands help you:

Query data with natural language
Generate code from descriptions
Analyze schemas for issues
Plan complex tasks automatically

aksara ai --help

ai query¶

Query your database using natural language.

aksara ai query "Users who signed up this week"

Output:

📊 Query: Users who signed up this week
📝 SQL: SELECT * FROM users WHERE created_at >= '2024-01-08'

Found 23 results:

| id       | email              | name     | created_at          |
|----------|-------------------|----------|---------------------|
| abc-123  | john@example.com  | John     | 2024-01-10 10:30:00 |
| def-456  | jane@example.com  | Jane     | 2024-01-09 14:22:00 |
...

Options¶

Option	Description
`--limit`, `-l`	Max results (default: 20)
`--output`, `-o`	Output file (csv, json)
`--sql-only`	Show SQL without executing
`--dry-run`	Parse query without executing
`--no-confirm`	Skip confirmation for updates

Examples¶

# Export to CSV
aksara ai query "All orders from last month" --output orders.csv

# Just show SQL
aksara ai query "Users with posts" --sql-only

# More results
aksara ai query "All products" --limit 100

ai generate¶

Generate code from natural language descriptions.

Generate Model¶

aksara ai generate model "BlogPost with title, slug, content, author FK, tags M2M"

Output:

class BlogPost(Model):
    """A blog post."""

    title = fields.String(max_length=200)
    slug = fields.String(max_length=200, unique=True)
    content = fields.Text()
    author = fields.ForeignKey("User", on_delete=fields.CASCADE)
    tags = fields.ManyToMany("Tag", related_name="posts")

    class Meta:
        table_name = "blog_posts"

Generate ViewSet¶

aksara ai generate viewset BlogPost

Generate Serializer¶

aksara ai generate serializer BlogPost

Generate Test¶

aksara ai generate test BlogPost --type crud

Generate CRUD¶

Generate model, serializer, viewset, and tests:

aksara ai generate crud "Product with name, price, description, category FK"

Options¶

Option	Description
`--output`, `-o`	Output file
`--append`	Append to existing file
`--dry-run`	Show without saving

ai doctor¶

Analyze your schema for issues and suggest fixes.

aksara ai doctor

Output:

🔍 Analyzing schema...

✓ 12 models analyzed
⚠️ 5 issues found

Issues:
  1. [HIGH] Post.author_id has no index (FK without index)
     Fix: Add db_index=True to ForeignKey

  2. [MEDIUM] User.email should be unique
     Fix: Add unique=True to Email field

  3. [LOW] Comment.created_at missing default
     Fix: Add auto_now_add=True

Run `aksara ai doctor --fix` to auto-fix issues.

Options¶

Option	Description
`--fix`	Auto-fix issues
`--fix-interactive`	Confirm each fix
`--min-severity`	Filter by severity
`--models`	Check specific models
`--output`, `-o`	Save report (json, html)

Examples¶

# Auto-fix all
aksara ai doctor --fix

# Interactive fixes
aksara ai doctor --fix-interactive

# Only high severity
aksara ai doctor --min-severity high

# Specific models
aksara ai doctor --models User,Post

# Save report
aksara ai doctor --output report.html

ai plan¶

Create a multi-step plan for complex tasks.

aksara ai plan "Add a commenting system to blog posts"

Output:

📋 Plan: Add commenting system

Step 1: Create Comment model
  Type: create_model
  File: models.py
  Dependencies: None

Step 2: Create CommentSerializer
  Type: create_serializer
  File: serializers.py
  Dependencies: Step 1

Step 3: Create CommentViewSet
  Type: create_viewset
  File: viewsets.py
  Dependencies: Steps 1, 2

Step 4: Add nested comments to PostSerializer
  Type: modify_file
  File: serializers.py
  Dependencies: Steps 1, 2

Step 5: Generate migration
  Type: create_migration
  Dependencies: Step 1

Step 6: Create tests
  Type: create_test
  File: tests/test_comments.py
  Dependencies: Steps 1-4

Execute plan? [y/N/i(nteractive)]

Options¶

Option	Description
`--execute`	Execute immediately
`--interactive`	Confirm each step
`--dry-run`	Preview changes only
`--save`	Save plan to file
`--load`	Load plan from file

Examples¶

# Execute immediately
aksara ai plan "Add user profiles" --execute

# Step by step
aksara ai plan "Add ratings" --interactive

# Save for later
aksara ai plan "Add search" --save search-plan.json

# Load and execute
aksara ai plan --load search-plan.json --execute

ai agent¶

Run an AI agent for complex autonomous tasks.

aksara ai agent "Review the codebase and suggest improvements"

Output:

🤖 Starting AI agent...

[Thinking] Analyzing codebase structure...
[Action] Reading models.py
[Action] Reading viewsets.py
[Thinking] Found potential issues...

Agent Report:
=============

1. N+1 Query in PostViewSet.list()
   File: viewsets.py:25
   Fix: Add select_related("author")

2. Missing index on Post.created_at
   File: models.py:15
   Fix: Add db_index=True

3. Duplicate code in serializers
   Files: serializers.py:10, serializers.py:45
   Fix: Extract common fields to base class

Would you like me to fix these issues? [y/N]

Options¶

Option	Description
`--interactive`	Confirm each action
`--tools`	Restrict available tools
`--timeout`	Max execution time
`--max-iterations`	Max think-act cycles

Examples¶

# Interactive mode
aksara ai agent "Improve test coverage" --interactive

# Read-only analysis
aksara ai agent "Review code" --tools query_records,list_models,read_file

# With timeout
aksara ai agent "Refactor models" --timeout 300

ai patch¶

Modify existing files with AI assistance.

aksara ai patch models.py "Add phone field to User model"

Output:

📝 Patch: Add phone field to User model

Preview:
--- models.py (original)
+++ models.py (modified)
@@ -10,6 +10,7 @@
 class User(Model):
     email = fields.Email(unique=True)
+    phone = fields.String(max_length=20, null=True)
     name = fields.String(max_length=100)

Apply patch? [y/N]

Options¶

Option	Description
`--dry-run`	Preview only
`--no-backup`	Don't create backup
`--force`	Skip confirmation

ai ask¶

Ask questions about your codebase.

aksara ai ask "How do I add pagination to a ViewSet?"

Output:

To add pagination to a ViewSet, set the pagination_class:

```python
from aksara.api import ModelViewSet, PageNumberPagination

class PostViewSet(ModelViewSet):
    model = Post
    pagination_class = PageNumberPagination
    page_size = 20

Based on your codebase, I see you're already using PageNumberPagination in UserViewSet (viewsets.py:15). You can follow the same pattern.

Related files: - viewsets.py:15 (existing pagination example) - settings.py:23 (default pagination settings)

---

## ai config

Configure AI settings.

```bash
aksara ai config --show

Output:

AI Configuration:
  Provider: openai
  Model: gpt-4
  API Key: sk-**** (configured)

Settings:
  Require confirmation: true
  Read-only mode: false
  Audit logging: true

Set Configuration¶

aksara ai config set AI_PROVIDER anthropic
aksara ai config set AI_MODEL claude-3-opus

Common Patterns¶

Generate and Apply¶

# Generate model
aksara ai generate model "Comment with text, author, post" > comment.py

# Review and add to models.py
cat comment.py >> models.py

# Generate migration
aksara makemigrations

Analyze and Fix¶

# Check for issues
aksara ai doctor

# Fix interactively
aksara ai doctor --fix-interactive

# Verify fixes
aksara test

Query and Export¶

# Query data
aksara ai query "Orders over $100 this month" --output big_orders.csv

# Analyze
aksara ai query "Average order value by customer" --output customer_value.json

Environment Variables¶

Variable	Description
`AKSARA_AI_API_KEY`	API key for AI provider
`AKSARA_AI_PROVIDER`	AI provider (openai, anthropic)
`AKSARA_AI_MODEL`	Model to use

AI Commands¶

Overview¶

ai query¶

Options¶

Examples¶

ai generate¶

Generate Model¶

Generate ViewSet¶

Generate Serializer¶

Generate Test¶

Generate CRUD¶

Options¶

ai doctor¶

Options¶

Examples¶

ai plan¶

Options¶

Examples¶

ai agent¶

Options¶

Examples¶

ai patch¶

Options¶

ai ask¶

Set Configuration¶

Common Patterns¶

Generate and Apply¶

Analyze and Fix¶

Query and Export¶

Environment Variables¶

Related Documentation¶