ATLAS — Deterministic Incident Analysis & Response (CLI)

ATLAS is a pip-installable CLI tool for deterministic incident analysis, root cause analysis (RCA), and response workflows.
It ingests real production logs and produces explainable, auditable outputs — without AI / ML.

Audience: SREs, DevOps, Platform engineers
Guarantee: Same input → same output. No black boxes.

---

Why ATLAS?

Most incident tools rely on opaque AI or heuristics that cannot be trusted during outages.
ATLAS is built for predictability, explainability, and production realism.

• 🔒 Zero-AI (no LLMs, no ML, no embeddings)
• 🧠 Fully deterministic
• 🧾 Rule-based & explainable
• 🧪 100% testable core
• ⚙️ Real CLI workflow (files in → decisions out)

---

What ATLAS Does

ATLAS v1 — Incident Analysis

1. Parses raw logs
2. Normalizes into a canonical incident schema
3. Classifies incident type
4. Infers probable root cause
5. Suggests remediation actions
6. Persists incident history (JSONL)

ATLAS v2 — Priority & Response Engine

ATLAS v2 converts analyzed incidents into decisions.

• Priority scoring (P0–P3)
• Rule-based decisions (IF–THEN)
• Workflow execution (step-by-step)
• State tracking (no infinite loops)
• Full audit trail
• Still zero-AI and deterministic

v1 answers “What happened?”
v2 answers “What should happen next?”

---

High-Level Architecture

flowchart TD
    A[Log File] --> B[Parsing]
    B --> C[Normalization]
    C --> D[Classification]
    D --> E[RCA]
    E --> F[Incident Schema]
    F --> G[Priority Engine]
    G --> H[Rule Engine]
    H --> I[Workflow Engine]
    I --> J[Actions + Audit Trail]

---

Install (PyPI)

pip install atlas-incident

Current version: 0.1.2

Usage

python -m atlas <log_file> <service> <environment>

Example:

python -m atlas prod_payments.log payments-api prod

What happens on run

1. v1 analyzes logs → Incident

2. v2 assigns priority → runs response workflow

Example Output

{
  "service": "payments-api",
  "environment": "prod",
  "severity": "LOW",
  "category": "UNKNOWN",
  "root_cause": "Unknown root cause",
  "actions": ["Escalate to on-call engineer"],
  "confidence": 0.6
}

=== ATLAS v2 RESULT ===
Priority: P2
Actions: []

Incident History

Each run is appended to incident_history.jsonl.

• Append-only
• Audit-friendly
• Replayable

Testing

pytest
pytest --cov=atlas_v2

• Deterministic tests
• Edge-case coverage
• CI-ready

Design Principles

• Determinism over intelligence
• Rules over models
• Schemas over guesses
• Clarity over cleverness
• Production realism over demos

---

License

This project is licensed under the MIT License, which permits use, modification, and distribution for both personal and commercial purposes.

ATLAS is provided “as is”, without warranty of any kind. It is suitable for learning, internal tooling, and experimental production use where deterministic and explainable behavior is required.