AI Integration

AI Integration Guide

Complete documentation for integrating with Cortex Linux's built-in Sapiens 0.27B reasoning engine.

---

Table of Contents

1. Overview
2. Command-Line Interface
3. HTTP API
4. Python Integration
5. Bash Integration
6. Performance Characteristics
7. Advanced Configuration

---

Overview

Cortex Linux integrates the Sapiens 0.27B reasoning engine directly into the operating system. The engine is accessible through multiple interfaces:

• CLI: cortex-ai command-line tool
• HTTP API: RESTful endpoints on port 8080
• System Services: Background services for automated tasks
• Library APIs: Python and other language bindings

Engine Specifications

• Model: Sapiens 0.27B (270 million parameters)
• Memory Usage: ~200MB RAM
• Latency: 50-200ms for typical queries
• Throughput: 5-10 queries/second (depending on hardware)
• Capabilities: Reasoning, planning, debugging, optimization

---

Command-Line Interface

Basic Usage

# Check version
cortex-ai --version

# Get help
cortex-ai --help

# Check service status
cortex-ai status

Core Commands

Reason

Perform general reasoning on a query.

# Basic reasoning
cortex-ai reason "What are the security implications of running a service as root?"

# With context
cortex-ai reason "Analyze this error: $ERROR_MESSAGE" --context "Application: web server, Language: Python"

# Save output to file
cortex-ai reason "Explain quantum computing" --output analysis.md

# Verbose output
cortex-ai reason "Debug this issue" --verbose

Example Output:

Query: What are the security implications of running a service as root?

Analysis:
Running services as root presents significant security risks:

1. Privilege Escalation: If the service is compromised, attackers gain full system access
2. Reduced Attack Surface: Root processes can modify critical system files
3. Principle of Least Privilege: Violates security best practices
4. Audit Trail: Root actions are harder to track and attribute

Recommendations:
- Create dedicated service user with minimal required permissions
- Use systemd user services or containerization
- Implement proper file permissions and SELinux/AppArmor policies
- Monitor root process execution with auditd

Confidence: 0.87
Processing time: 145ms

Plan

Generate structured plans for tasks or projects.

# Create a deployment plan
cortex-ai plan "Deploy a Python web application to production" \
  --requirements "High availability, zero downtime, monitoring"

# Plan with existing configuration
cortex-ai plan "Optimize database performance" \
  --config /etc/postgresql/postgresql.conf \
  --output plan.json

# Multi-step planning
cortex-ai plan "Migrate from MySQL to PostgreSQL" \
  --steps 10 \
  --format markdown

Example Output:

# Deployment Plan: Python Web Application

## Phase 1: Preparation (Days 1-2)
1. Review application requirements
2. Set up staging environment
3. Configure CI/CD pipeline
4. Prepare database migrations

## Phase 2: Infrastructure (Days 3-5)
1. Provision load balancer
2. Configure multiple application servers
3. Set up database replication
4. Configure monitoring and logging

## Phase 3: Deployment (Days 6-7)
1. Deploy to staging
2. Run integration tests
3. Blue-green deployment to production
4. Verify zero-downtime transition

## Phase 4: Validation (Day 8)
1. Monitor application metrics
2. Verify high availability
3. Load testing
4. Rollback plan ready

Risk Assessment:
- Medium: Database migration complexity
- Low: Application deployment
- Low: Infrastructure setup

Estimated Duration: 8 days
Confidence: 0.82

Debug

Analyze errors, logs, and system issues.

# Debug from log file
cortex-ai debug /var/log/app/error.log

# Debug with stack trace
cortex-ai debug --stack-trace stacktrace.txt

# Debug system issue
cortex-ai debug "High CPU usage on server" \
  --metrics /var/log/system-metrics.json

# Interactive debugging
cortex-ai debug --interactive

Example Output:

Error Analysis: /var/log/app/error.log

Primary Issue: Database connection timeout
Location: app/database.py:142
Frequency: 15 occurrences in last hour

Root Cause Analysis:
1. Connection pool exhausted (max_connections: 10, active: 10)
2. Long-running queries blocking connection release
3. Missing connection timeout configuration

Suggested Fixes:
1. Increase connection pool size:
   MAX_CONNECTIONS=20

2. Add connection timeout:
   CONNECTION_TIMEOUT=30

3. Implement query timeout:
   QUERY_TIMEOUT=10

4. Add connection retry logic with exponential backoff

Code Changes Required:
- app/database.py: Update connection pool configuration
- app/models.py: Add query timeout decorators
- app/utils.py: Implement retry mechanism

Confidence: 0.91
Processing time: 178ms

Optimize

Optimize code, configuration, or system performance.

# Optimize code file
cortex-ai optimize /path/to/script.py --language python

# Optimize configuration
cortex-ai optimize /etc/nginx/nginx.conf --target performance

# Optimize system
cortex-ai optimize --target system --output recommendations.json

# Optimize with constraints
cortex-ai optimize "Database query" \
  --constraints "Must use existing indexes, no schema changes"

Example Output:

Optimization Analysis: /path/to/script.py

Current Performance:
- Execution time: 2.3s
- Memory usage: 45MB
- CPU utilization: 85%

Optimization Opportunities:

1. Database Query Optimization (High Impact)
   - Replace N+1 queries with JOIN
   - Add database indexes on frequently queried columns
   - Estimated improvement: 60% faster

2. Caching Strategy (Medium Impact)
   - Implement Redis caching for user data
   - Cache expensive computations
   - Estimated improvement: 40% faster

3. Algorithm Improvements (Medium Impact)
   - Replace O(n²) sorting with O(n log n)
   - Use generators instead of lists for large datasets
   - Estimated improvement: 30% faster

4. Code Refactoring (Low Impact)
   - Remove redundant function calls
   - Optimize string concatenation
   - Estimated improvement: 10% faster

Recommended Changes:
[Detailed code suggestions with line numbers]

Expected Results:
- Execution time: 0.7s (70% improvement)
- Memory usage: 32MB (29% reduction)
- CPU utilization: 55% (35% reduction)

Confidence: 0.85
Processing time: 203ms

Advanced CLI Options

# Set model parameters
cortex-ai reason "Query" \
  --temperature 0.7 \
  --max-tokens 500 \
  --top-p 0.9

# Use specific model version
cortex-ai reason "Query" --model sapiens-0.27b-v2

# Enable streaming output
cortex-ai reason "Long query" --stream

# Set timeout
cortex-ai reason "Query" --timeout 30

# Use custom config file
cortex-ai reason "Query" --config /path/to/config.yaml

---

HTTP API

The Cortex AI HTTP API provides RESTful endpoints for programmatic access.

Base URL

• Local: http://localhost:8080
• Network: http://your-server-ip:8080

Authentication

By default, the API runs without authentication. For production, enable API keys:

# /etc/cortex-ai/config.yaml
security:
  api_key_required: true
  api_key: your-secret-key-here

Then include in requests:

curl -H "X-API-Key: your-secret-key-here" ...

Endpoints

POST /reason

Perform reasoning on a query.

Request:

curl -X POST http://localhost:8080/reason \
  -H "Content-Type: application/json" \
  -d '{
    "query": "What are best practices for securing a REST API?",
    "context": "Python Flask application",
    "temperature": 0.7,
    "max_tokens": 500
  }'

Response:

{
  "result": "Best practices for securing a REST API include:\n\n1. Authentication: Use JWT tokens or OAuth 2.0\n2. HTTPS: Always use TLS encryption\n3. Input validation: Sanitize all user inputs\n4. Rate limiting: Prevent abuse and DDoS attacks\n5. CORS: Configure Cross-Origin Resource Sharing properly\n6. Error handling: Don't expose sensitive information in errors",
  "confidence": 0.89,
  "processing_time_ms": 156,
  "tokens_used": 127,
  "model_version": "sapiens-0.27b"
}

POST /plan

Generate structured plans.

Request:

curl -X POST http://localhost:8080/plan \
  -H "Content-Type: application/json" \
  -d '{
    "task": "Set up CI/CD pipeline for microservices",
    "requirements": ["GitHub Actions", "Docker", "Kubernetes"],
    "format": "markdown",
    "steps": 8
  }'

Response:

{
  "plan": "# CI/CD Pipeline Setup Plan\n\n## Phase 1: Repository Setup\n...",
  "estimated_duration": "2 weeks",
  "risk_level": "medium",
  "confidence": 0.84,
  "processing_time_ms": 234
}

POST /debug

Debug errors and issues.

Request:

curl -X POST http://localhost:8080/debug \
  -H "Content-Type: application/json" \
  -d '{
    "error_log": "Traceback (most recent call last):\n  File \"app.py\", line 42, in <module>\n    result = process_data(data)\nKeyError: 'missing_key'",
    "context": {
      "language": "Python",
      "framework": "Flask",
      "environment": "production"
    }
  }'

Response:

{
  "analysis": {
    "error_type": "KeyError",
    "location": "app.py:42",
    "root_cause": "Accessing dictionary key that doesn't exist",
    "severity": "medium"
  },
  "suggestions": [
    "Use dict.get() with default value",
    "Check key existence with 'in' operator",
    "Validate input data before processing"
  ],
  "code_fixes": [
    {
      "file": "app.py",
      "line": 42,
      "current": "result = process_data(data)",
      "suggested": "result = process_data(data) if 'required_key' in data else default_value"
    }
  ],
  "confidence": 0.92,
  "processing_time_ms": 167
}

POST /optimize

Optimize code or configuration.

Request:

curl -X POST http://localhost:8080/optimize \
  -H "Content-Type: application/json" \
  -d '{
    "target": "code",
    "content": "def process_items(items):\n    result = []\n    for item in items:\n        if item.valid:\n            result.append(item.process())\n    return result",
    "language": "python",
    "goals": ["performance", "readability"]
  }'

Response:

{
  "optimizations": [
    {
      "type": "list_comprehension",
      "description": "Replace loop with list comprehension",
      "current_code": "def process_items(items):\n    result = []\n    for item in items:\n        if item.valid:\n            result.append(item.process())\n    return result",
      "optimized_code": "def process_items(items):\n    return [item.process() for item in items if item.valid]",
      "improvement_estimate": "15% faster, more Pythonic"
    }
  ],
  "overall_improvement": "15-20% performance gain",
  "confidence": 0.88,
  "processing_time_ms": 189
}

GET /health

Check API health status.

Request:

curl http://localhost:8080/health

Response:

{
  "status": "healthy",
  "version": "1.0.0",
  "model_loaded": true,
  "memory_usage_mb": 187,
  "uptime_seconds": 3600,
  "requests_processed": 1234
}

GET /status

Get detailed system status.

Request:

curl http://localhost:8080/status

Response:

{
  "service": "cortex-ai",
  "version": "1.0.0",
  "model": {
    "name": "sapiens-0.27b",
    "version": "1.0",
    "loaded": true,
    "memory_mb": 200
  },
  "system": {
    "cpu_usage_percent": 45.2,
    "memory_usage_mb": 187,
    "disk_usage_percent": 23.1
  },
  "performance": {
    "avg_response_time_ms": 156,
    "requests_per_second": 6.2,
    "total_requests": 1234
  }
}

Error Responses

All endpoints return standard HTTP status codes:

• 200 OK: Successful request
• 400 Bad Request: Invalid request parameters
• 401 Unauthorized: Missing or invalid API key
• 429 Too Many Requests: Rate limit exceeded
• 500 Internal Server Error: Server error

Error response format:

{
  "error": {
    "code": "INVALID_QUERY",
    "message": "Query cannot be empty",
    "details": {}
  }
}

---

Python Integration

Installation

# Install Python SDK
pip install cortex-ai

# Or from source
git clone https://github.com/cortexlinux/cortex-python-sdk
cd cortex-python-sdk
pip install -e .

Basic Usage

from cortex import AI

# Initialize client
ai = AI()

# Or with custom configuration
ai = AI(
    host='localhost',
    port=8080,
    api_key='your-key'  # Optional
)

# Reason
result = ai.reason("What is the difference between REST and GraphQL?")
print(result.result)
print(f"Confidence: {result.confidence}")

# Plan
plan = ai.plan(
    task="Deploy application to Kubernetes",
    requirements=["High availability", "Auto-scaling"],
    format="markdown"
)
print(plan.plan)

# Debug
debug_result = ai.debug(
    error_log="Error: Connection refused",
    context={"application": "web server", "language": "Python"}
)
for suggestion in debug_result.suggestions:
    print(f"- {suggestion}")

# Optimize
optimization = ai.optimize(
    code="""
def slow_function(data):
    result = []
    for item in data:
        if item > 0:
            result.append(item * 2)
    return result
    """,
    language="python",
    goals=["performance"]
)
print(optimization.optimizations[0].optimized_code)

Advanced Usage

from cortex import AI
import asyncio

# Async client
async def main():
    ai = AI()
    
    # Async reasoning
    result = await ai.reason_async(
        "Analyze this architecture",
        temperature=0.7,
        max_tokens=1000
    )
    print(result.result)

# Streaming
ai = AI()
for chunk in ai.reason_stream("Long query that generates streaming output"):
    print(chunk, end='', flush=True)

# Batch processing
queries = [
    "What is microservices architecture?",
    "Explain container orchestration",
    "Describe CI/CD best practices"
]

results = ai.reason_batch(queries)
for query, result in zip(queries, results):
    print(f"Q: {query}\nA: {result.result}\n")

Error Handling

from cortex import AI
from cortex.exceptions import CortexError, CortexTimeoutError

ai = AI()

try:
    result = ai.reason("Query", timeout=5)
except CortexTimeoutError:
    print("Request timed out")
except CortexError as e:
    print(f"Error: {e.message}")
except Exception as e:
    print(f"Unexpected error: {e}")

---

Bash Integration

Basic Script Integration

#!/bin/bash

# Function to query Cortex AI
cortex_reason() {
    local query="$1"
    curl -s -X POST http://localhost:8080/reason \
        -H "Content-Type: application/json" \
        -d "{\"query\": \"$query\"}" | \
        jq -r '.result'
}

# Usage
RESULT=$(cortex_reason "How do I check disk usage in Linux?")
echo "$RESULT"

Advanced Bash Integration

#!/bin/bash

# Configuration
CORTEX_HOST="${CORTEX_HOST:-localhost}"
CORTEX_PORT="${CORTEX_PORT:-8080}"
CORTEX_API_KEY="${CORTEX_API_KEY:-}"

# Helper function
cortex_api() {
    local endpoint="$1"
    local data="$2"
    local headers=(-H "Content-Type: application/json")
    
    if [ -n "$CORTEX_API_KEY" ]; then
        headers+=(-H "X-API-Key: $CORTEX_API_KEY")
    fi
    
    curl -s -X POST "http://${CORTEX_HOST}:${CORTEX_PORT}${endpoint}" \
        "${headers[@]}" \
        -d "$data"
}

# Reason wrapper
cortex_reason() {
    local query="$1"
    local context="${2:-}"
    local json_data="{\"query\": \"$query\""
    
    if [ -n "$context" ]; then
        json_data+=", \"context\": \"$context\""
    fi
    json_data+="}"
    
    cortex_api "/reason" "$json_data" | jq -r '.result'
}

# Debug wrapper
cortex_debug() {
    local error_log="$1"
    local log_content
    
    if [ -f "$error_log" ]; then
        log_content=$(cat "$error_log")
    else
        log_content="$error_log"
    fi
    
    cortex_api "/debug" "{\"error_log\": $(jq -Rs . <<< "$log_content")}" | \
        jq -r '.suggestions[]'
}

# Example usage
if [ "$1" = "debug" ]; then
    cortex_debug "$2"
elif [ "$1" = "reason" ]; then
    cortex_reason "$2" "$3"
else
    echo "Usage: $0 {debug|reason} <input> [context]"
fi

System Integration Examples

# Auto-debug on error
trap 'cortex_debug "$(tail -50 /var/log/app/error.log)"' ERR

# Optimize configuration on change
inotifywait -m /etc/nginx/nginx.conf | while read event; do
    cortex_optimize /etc/nginx/nginx.conf
done

# Monitor and reason about system metrics
while true; do
    METRICS=$(collect_metrics)
    ANALYSIS=$(cortex_reason "Analyze these system metrics: $METRICS")
    echo "$ANALYSIS" >> /var/log/system-analysis.log
    sleep 300
done

---

Performance Characteristics

Latency

Typical response times for different query types:

Query Type	Average Latency	P95 Latency	P99 Latency
Simple reasoning	50-100ms	150ms	200ms
Complex reasoning	100-200ms	300ms	500ms
Planning	150-250ms	400ms	600ms
Debugging	100-180ms	280ms	450ms
Optimization	180-300ms	500ms	800ms

Throughput

• Single-threaded: 5-7 queries/second
• Multi-threaded (4 cores): 15-20 queries/second
• Concurrent requests: Up to 50 simultaneous requests (with queuing)

Resource Usage

• Memory: ~200MB base + ~10MB per active request
• CPU: 15-30% per active request (single core)
• Disk I/O: Minimal (model loaded in memory)

Optimization Tips

# /etc/cortex-ai/config.yaml
ai:
  # Increase threads for multi-core systems
  num_threads: 4
  
  # Adjust memory allocation
  max_memory_mb: 512
  
  # Enable request batching
  batch_size: 10
  
  # Cache frequent queries
  enable_caching: true
  cache_ttl_seconds: 3600

---

Advanced Configuration

Custom Model Parameters

Edit /etc/cortex-ai/config.yaml:

ai:
  model_path: /usr/lib/cortex-ai/models/sapiens-0.27b
  
  # Generation parameters
  temperature: 0.7        # Creativity (0.0-1.0)
  top_p: 0.9             # Nucleus sampling
  top_k: 40              # Top-k sampling
  max_tokens: 512        # Maximum output length
  
  # Performance
  num_threads: 4         # CPU threads
  max_memory_mb: 512     # Memory limit
  batch_size: 1          # Batch processing
  
  # Caching
  enable_caching: true
  cache_ttl_seconds: 3600
  cache_max_size_mb: 100

Logging Configuration

logging:
  level: INFO  # DEBUG, INFO, WARNING, ERROR
  file: /var/log/cortex-ai/cortex-ai.log
  max_size_mb: 100
  backup_count: 5
  format: json  # json or text

Rate Limiting

rate_limiting:
  enabled: true
  requests_per_minute: 60
  requests_per_hour: 1000
  burst_size: 10

---

Next Steps

• Explore Use Cases: Use Cases and Tutorials
• Developer Resources: Developer Documentation
• Performance Tuning: Technical Specifications

---

Last updated: 2024