AI On-Call Agent

AI agents that replace human on-call engineers. Analyze errors, identify root causes, and generate code fixes automatically.

---

Overview

NoMoreOnCall is an automated error analysis and debugging system that helps developers quickly identify and fix application errors. It combines error tracing, code context analysis, and automated code fix suggestions—reducing mean-time-to-resolution (MTTR) for production incidents.

Key Features

Feature	Description
Error Analysis	Analyzes errors and generates detailed JSON reports with root cause analysis
Code Context	Provides code-level context, blame information, and stack traces
Automated Fixes	Generates code change suggestions for 15+ error types
Notification API	Integrated API for error alerting and automation workflows
Git Integration	Creates fix branches, patches, and pull requests automatically

---

Quick Start

# 1. Clone and setup
git clone https://github.com/hanu-tayal/ai-oncall-agent.git
cd ai-oncall-agent
python -m venv .venv
.venv\Scripts�ctivate   # Windows
# source .venv/bin/activate  # macOS/Linux

# 2. Install dependencies
pip install -r requirements.txt

# 3. Run the demo
python demo.py

The demo will analyze a sample database timeout error (ERR_123) and generate fix suggestions.

---

Architecture

Error Occurs → Debug Analyzer → Code Context & Blame → Root Cause Analysis
                    ↓
              Issue JSON File
                    ↓
              Code Fixer → Suggested Changes → Patch/PR

See ARCHITECTURE.md for a detailed technical overview.

---

Components

1. Debug Analyzer (debug_analyzer_v2.py)

Fetches error details (from API or mock data), analyzes related code files with context and blame information, and generates structured JSON reports.

python debug_analyzer_v2.py ERR_123

2. Code Fixer (code_fixer.py)

Reads the issue JSON, suggests specific code changes based on error type, and can apply changes, create patches, and open pull requests.

python code_fixer.py issue_ERR_123.json

3. Notification API (integrated in demo.py)

Receives error notifications from monitoring systems. Runs in-process during the demo on port 8001.

---

Supported Error Types

Error Type	Fix Strategy
DatabaseError	Increase timeouts, parameterize queries, add pooling
AuthenticationError	Improve token validation, better header extraction
ConnectionError	Add retry logic
TimeoutError	Increase/double timeouts
ValidationError	Add comprehensive input validation
ResourceNotFoundError	Add default values to lookups
PermissionError	Add role-based permission checks
RateLimitError	Add rate limiting decorators
MemoryError	Use generators, memory-efficient structures
ConcurrencyError	Add thread safety, locks
ConfigurationError	Add default values, validate env vars
SecurityError	Add password hashing, secure token handling
NetworkError	Add timeouts/retries
FileSystemError	Use context managers, check file existence
SerializationError	Safe JSON/pickle serialization

---

Configuration

Create a .env file for custom settings:

DEBUG_API_BASE_URL=http://localhost:8000
DEBUG_API_KEY=dummy_key
REPO_PATH=.
GITHUB_TOKEN=your_token        # For PR creation
GITHUB_REPO=username/repo     # For PR creation

---

Project Structure

NoMoreOnCall/
├── debug_analyzer_v2.py   # Main error analyzer
├── code_fixer.py          # Code fix suggestions & Git/PR integration
├── demo.py                # End-to-end demo script
├── ARCHITECTURE.md        # Technical architecture
├── requirements.txt
└── data/                  # Sample data

---

Documentation

• ARCHITECTURE.md — Technical architecture and system flow
• CONTRIBUTING.md — How to contribute

---

License

MIT