🐛 DebugAI

AI-powered CLI tool that turns any stack trace into a root cause analysis — in seconds.

No dashboard. No account. No SDK to integrate. Just pipe in a log and get answers.

cat production.log | debuai --ai

---

Demo

https://github.com/user-attachments/assets/5771164c-865e-4c95-aff8-3cc7e2dae0f2

---

Install

pip install debuai

---

Quickstart

# Analyse a log file
debuai error.log

# Pipe from any source
cat error.log | debuai
kubectl logs my-pod | debuai
docker logs my-container | debuai

# With AI analysis
debuai error.log --ai

# Machine-readable JSON output
debuai error.log --json
debuai error.log --ai --json | jq .

# Choose your AI provider
debuai error.log --ai --provider openai
debuai error.log --ai --provider anthropic
debuai error.log --ai --provider nvidia

# Analyse the last N errors in a log
debuai error.log --top 3

---

What It Does

Given a stack trace like this:

Traceback (most recent call last):
  File "server.py", line 10, in handle_request
    result = process_data(payload)
  File "processor.py", line 25, in process_data
    value = parse_input(raw)
  File "parser.py", line 8, in parse_input
    return int(raw)
ValueError: invalid literal for int() with base 10: 'abc'

DebugAI gives you this:

🔥 Exception Type   ValueError
📍 Failure Origin   parse_input
🔗 Execution Chain  handle_request → process_data → parse_input

🤖 AI Root Cause    The function receives a string that cannot be
                    converted to int — 'abc' is passed where a
                    numeric string is expected.

🛠 Suggested Fix    Validate input before conversion:
                    if not raw.isdigit(): raise ValueError(...)

🛡 Prevention       Add input validation at the API boundary so
                    invalid types never reach the parsing layer.

🟢 Confidence: HIGH

---

Supported Languages

Language	Detection	Parser
Python	Traceback (most recent call last)	✅
Java	.java: in frames	✅
Go	goroutine / panic:	✅
C# / .NET	at Namespace.Class.Method()	✅
Node.js	.js: in frames	✅

Language is auto-detected — no --lang flag needed.

---

AI Providers

DebugAI works with whichever key you already have. Set one environment variable and it just works.

export OPENAI_API_KEY=...       # Uses gpt-4o-mini
export ANTHROPIC_API_KEY=...    # Uses claude-haiku
export NVIDIA_API_KEY=...       # Uses NVIDIA inference

Auto-detection order: OpenAI → Anthropic → NVIDIA. Override with --provider.

---

JSON Output

The --json flag outputs clean, pipeable JSON — no Rich formatting, no colour codes.

debuai error.log --ai --json

{
  "exception": "ValueError",
  "failure_origin": "parse_input",
  "execution_chain": ["handle_request", "process_data", "parse_input"],
  "source_file": "parser.py",
  "language": "python",
  "ai": {
    "root_cause": "...",
    "fix": "...",
    "prevention": "...",
    "confidence": "high"
  }
}

Pipe into jq, Slack bots, Jira integrations, or CI pipelines.

---

CI / CD Integration

Add DebugAI to your GitHub Actions workflow to get AI analysis on every failed build:

- name: Analyse failure
  if: failure()
  run: |
    pip install debuai
    cat logs/error.log | debuai --ai --json
  env:
    OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}

---

Setup for Development

git clone https://github.com/kanhaiya-bhayana/debugai.git
cd debugai
python -m venv venv
source venv/bin/activate        # Mac/Linux
venv\Scripts\activate           # Windows
pip install -e ".[dev]"

Run tests:

python -m pytest tests/ -v

Set your API key:

export OPENAI_API_KEY=your_key_here

---

Project Structure

debugai/
├── parser/
│   ├── python.py        # Python traceback parser
│   ├── java.py          # Java / Spring parser
│   ├── go.py            # Go panic parser
│   ├── csharp.py        # C# / .NET parser
│   ├── node.py          # Node.js parser
│   └── registry.py      # Language auto-detection router
├── providers/
│   ├── openai_provider.py
│   ├── anthropic_provider.py
│   └── nvidia.py
├── analyzer.py          # Core extraction logic
├── ai_analyzer.py       # Provider selection + AI orchestration
└── cli.py               # Typer CLI entry point
tests/
├── test_parsers.py      # 50 parser tests
├── test_analyzer.py     # 28 analyzer tests
└── test_safeguards.py   # 13 edge case + safeguard tests

---

Roadmap

• Python, Java, Go, C#, Node.js parsers
• Multi-provider AI backend (OpenAI, Anthropic, NVIDIA)
• Structured JSON output
• CI/CD pipeline
• PyPI publish
• GitHub issue search — link traces to known issues automatically
• Kubernetes / Docker log stream support
• VS Code extension
• Web UI for team sharing

---

Contributing

Contributions are welcome. Please read CONTRIBUTING.md before opening a PR.

# Run the full test suite before submitting
python -m pytest tests/ -v

---

License

MIT — see LICENSE for details.

---

Built with ❤️ for developers who are tired of Googling stack traces.