RomanNum3ral
/
cai
mirror of https://github.com/aliasrobotics/cai.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
cai/Cybersecuritypythonrubyproject

563 lines
18 KiB
Plaintext
Raw Blame History

# Project: Docker Cyber Analyzer (AI-ish rules + baseline checks)
A minimal, extensible project that inspects your local Docker environment for risk signals often present in cybercrime/abuse investigations (e.g., privileged containers, root users, broad port exposure, suspicious secrets in env). Results are stored in SQLite and are accessible via a Python FastAPI and a Ruby Sinatra dashboard.
> **Note**: You can enrich the Python analyzer with ML/AI later (e.g., scoring findings, anomaly detection, or rule-learning from labeled cases). The current build uses a deterministic ruleset that is easy to extend.
---
## File tree
```
/docker-cyber-analyzer
├─ README.md
├─ docker-compose.yml
├─ data/ # SQLite file lives here (shared volume)
├─ python_service/
│ ├─ requirements.txt
│ ├─ Dockerfile
│ ├─ database.py
│ ├─ models.py
│ ├─ analyzer.py # Core checks
│ └─ app.py # FastAPI server
└─ ruby_dashboard/
├─ Gemfile
├─ Dockerfile
└─ app.rb # Sinatra mini UI
```
---
## Quick start (Docker Compose)
```bash
# In project root
docker compose up --build
# Trigger a scan
curl -X POST http://localhost:8000/analyze
# Explore results
# API docs: http://localhost:8000/docs
# Dashboard: http://localhost:4567
```
---
## Python service (FastAPI)
### `python_service/requirements.txt`
```txt
fastapi==0.115.0
uvicorn[standard]==0.30.6
pydantic==2.9.2
SQLAlchemy==2.0.35
docker==7.1.0
```
### `python_service/Dockerfile`
```dockerfile
FROM python:3.11-slim
WORKDIR /app
COPY requirements.txt ./
RUN pip install --no-cache-dir -r requirements.txt
COPY . .
ENV PYTHONUNBUFFERED=1
CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "8000"]
```
### `python_service/database.py`
```python
import os
from sqlalchemy import create_engine
from sqlalchemy.orm import sessionmaker, DeclarativeBase
DB_PATH = os.getenv("DB_PATH", "/app/data/analyzer.db")
class Base(DeclarativeBase):
pass
engine = create_engine(f"sqlite:///{DB_PATH}", echo=False, future=True)
SessionLocal = sessionmaker(bind=engine, autoflush=False, autocommit=False)
```
### `python_service/models.py`
```python
from sqlalchemy import Column, Integer, String, Boolean, DateTime, ForeignKey, Text, Float
from sqlalchemy.orm import relationship, Mapped, mapped_column
from datetime import datetime
from database import Base
class Case(Base):
__tablename__ = "cases"
id: Mapped[int] = mapped_column(Integer, primary_key=True, index=True)
created_at: Mapped[datetime] = mapped_column(DateTime, default=datetime.utcnow)
note: Mapped[str] = mapped_column(String(255), default="auto-scan")
containers = relationship("ContainerReport", back_populates="case", cascade="all, delete-orphan")
images = relationship("ImageReport", back_populates="case", cascade="all, delete-orphan")
findings = relationship("Finding", back_populates="case", cascade="all, delete-orphan")
class ContainerReport(Base):
__tablename__ = "container_reports"
id: Mapped[int] = mapped_column(Integer, primary_key=True)
case_id: Mapped[int] = mapped_column(Integer, ForeignKey("cases.id"))
container_id: Mapped[str] = mapped_column(String(128))
name: Mapped[str] = mapped_column(String(256))
image: Mapped[str] = mapped_column(String(256))
privileged: Mapped[bool] = mapped_column(Boolean, default=False)
runs_as_root: Mapped[bool] = mapped_column(Boolean, default=False)
host_network: Mapped[bool] = mapped_column(Boolean, default=False)
port_0_0_0_0: Mapped[bool] = mapped_column(Boolean, default=False)
added_caps: Mapped[str] = mapped_column(Text, default="")
mounts: Mapped[str] = mapped_column(Text, default="")
suspicious_env: Mapped[str] = mapped_column(Text, default="")
case = relationship("Case", back_populates="containers")
class ImageReport(Base):
__tablename__ = "image_reports"
id: Mapped[int] = mapped_column(Integer, primary_key=True)
case_id: Mapped[int] = mapped_column(Integer, ForeignKey("cases.id"))
image_id: Mapped[str] = mapped_column(String(128))
repo_tags: Mapped[str] = mapped_column(Text, default="")
size_mb: Mapped[float] = mapped_column(Float, default=0.0)
latest_tag: Mapped[bool] = mapped_column(Boolean, default=False)
age_days: Mapped[float] = mapped_column(Float, default=0.0)
case = relationship("Case", back_populates="images")
class Finding(Base):
__tablename__ = "findings"
id: Mapped[int] = mapped_column(Integer, primary_key=True)
case_id: Mapped[int] = mapped_column(Integer, ForeignKey("cases.id"))
subject_type: Mapped[str] = mapped_column(String(32)) # 'container' | 'image' | 'system'
subject_id: Mapped[str] = mapped_column(String(128))
severity: Mapped[str] = mapped_column(String(16)) # LOW/MEDIUM/HIGH/CRITICAL
title: Mapped[str] = mapped_column(String(256))
detail: Mapped[str] = mapped_column(Text, default="")
case = relationship("Case", back_populates="findings")
```
### `python_service/analyzer.py`
```python
import os, re, datetime
from typing import List, Dict, Any
import docker
from sqlalchemy.orm import Session
from database import SessionLocal, engine
from models import Base, Case, ContainerReport, ImageReport, Finding
SUS_ENV_PATTERNS = [
r"AWS_(ACCESS_KEY|SECRET|SESSION)_?ID?", r"API_KEY", r"SECRET", r"TOKEN",
r"PASSWORD", r"PRIVATE_KEY", r"GCP_|GOOGLE_APPLICATION_CREDENTIALS", r"AZURE_"
]
IMAGE_AGE_LIMIT_DAYS = int(os.getenv("ANALYZER_IMAGE_AGE_DAYS", "365"))
IMAGE_SIZE_LIMIT_MB = int(os.getenv("ANALYZER_IMAGE_SIZE_MB", "800"))
def init_db():
Base.metadata.create_all(bind=engine)
def matches_suspicious(key: str) -> bool:
for p in SUS_ENV_PATTERNS:
if re.search(p, key, flags=re.IGNORECASE):
return True
return False
def analyze() -> int:
init_db()
client = docker.from_env()
with SessionLocal() as db:
case = Case()
db.add(case)
db.commit()
db.refresh(case)
analyze_containers(client, db, case.id)
analyze_images(client, db, case.id)
db.commit()
return case.id
def analyze_containers(client, db: Session, case_id: int):
for c in client.containers.list(all=True):
attrs = c.attrs
cfg = attrs.get("Config", {}) or {}
host_cfg = attrs.get("HostConfig", {}) or {}
network_mode = host_cfg.get("NetworkMode")
privileged = bool(host_cfg.get("Privileged"))
added_caps = host_cfg.get("CapAdd") or []
user = cfg.get("User") or ""
runs_as_root = (user.strip() == "" or user.strip() == "0")
ports = attrs.get("NetworkSettings", {}).get("Ports") or {}
port_0_0_0_0 = any(bind for p in ports.values() if p for bind in p if bind.get("HostIp") in ("0.0.0.0", "::"))
host_network = (network_mode == "host")
mounts = attrs.get("Mounts") or []
mounts_str = "; ".join([f"{m.get('Source')} -> {m.get('Destination')}" for m in mounts])
env_list = cfg.get("Env") or []
env_keys = [e.split("=", 1)[0] for e in env_list]
sus_keys = [k for k in env_keys if matches_suspicious(k)]
suspicious_env = ", ".join(sus_keys)
report = ContainerReport(
case_id=case_id,
container_id=c.id[:12],
name=(attrs.get("Name") or c.name or "").lstrip("/"),
image=cfg.get("Image") or c.image.tags[0] if c.image.tags else c.image.short_id,
privileged=privileged,
runs_as_root=runs_as_root,
host_network=host_network,
port_0_0_0_0=port_0_0_0_0,
added_caps=",".join(added_caps),
mounts=mounts_str,
suspicious_env=suspicious_env,
)
db.add(report)
db.flush()
# Findings
if privileged:
add_finding(db, case_id, "container", report.container_id, "HIGH", "Privileged container", f"{report.name} runs with --privileged")
if runs_as_root:
add_finding(db, case_id, "container", report.container_id, "MEDIUM", "Runs as root", f"{report.name} has no non-root user configured")
if host_network:
add_finding(db, case_id, "container", report.container_id, "MEDIUM", "Host networking", f"{report.name} uses host network")
if port_0_0_0_0:
add_finding(db, case_id, "container", report.container_id, "MEDIUM", "Public port exposure", f"{report.name} exposes ports on 0.0.0.0")
if added_caps:
add_finding(db, case_id, "container", report.container_id, "LOW", "Extra capabilities", f"Capabilities added: {report.added_caps}")
if mounts_str:
add_finding(db, case_id, "container", report.container_id, "LOW", "Mounted host paths", f"Mounts: {mounts_str}")
if suspicious_env:
add_finding(db, case_id, "container", report.container_id, "HIGH", "Suspicious env keys", f"Keys: {suspicious_env}")
def analyze_images(client, db: Session, case_id: int):
now = datetime.datetime.utcnow()
for img in client.images.list():
tags = img.tags or []
repo_tags = ", ".join(tags)
latest_tag = any(t.endswith(":latest") for t in tags)
size_mb = round((getattr(img, 'attrs', {}).get('Size') or 0) / (1024*1024), 2)
created_str = getattr(img, 'attrs', {}).get('Created')
age_days = 0.0
if created_str:
try:
created = datetime.datetime.fromisoformat(created_str.replace('Z','+00:00'))
age_days = (now - created).total_seconds() / 86400.0
except Exception:
age_days = 0.0
report = ImageReport(
case_id=case_id,
image_id=img.short_id.replace("sha256:", ""),
repo_tags=repo_tags,
size_mb=size_mb,
latest_tag=latest_tag,
age_days=round(age_days, 1)
)
db.add(report)
db.flush()
if latest_tag:
add_finding(db, case_id, "image", report.image_id, "LOW", "Floating 'latest' tag", f"{repo_tags}")
if size_mb > IMAGE_SIZE_LIMIT_MB:
add_finding(db, case_id, "image", report.image_id, "LOW", "Large image size", f"{size_mb} MB > {IMAGE_SIZE_LIMIT_MB} MB threshold")
if age_days > IMAGE_AGE_LIMIT_DAYS:
add_finding(db, case_id, "image", report.image_id, "LOW", "Old image", f"{age_days:.0f} days old (> {IMAGE_AGE_LIMIT_DAYS})")
def add_finding(db: Session, case_id: int, subject_type: str, subject_id: str, severity: str, title: str, detail: str):
f = Finding(case_id=case_id, subject_type=subject_type, subject_id=subject_id, severity=severity, title=title, detail=detail)
db.add(f)
```
### `python_service/app.py`
```python
from fastapi import FastAPI
from pydantic import BaseModel
from typing import Optional
from database import SessionLocal
from models import Case, ContainerReport, ImageReport, Finding
from analyzer import analyze, init_db
app = FastAPI(title="Docker Cyber Analyzer API", version="0.1.0")
class AnalyzeRequest(BaseModel):
note: Optional[str] = None
@app.on_event("startup")
def on_startup():
init_db()
@app.get("/healthz")
def healthz():
return {"ok": True}
@app.post("/analyze")
def run_analyze(req: AnalyzeRequest | None = None):
case_id = analyze()
if req and req.note:
with SessionLocal() as db:
c = db.get(Case, case_id)
if c:
c.note = req.note
db.commit()
return {"case_id": case_id}
@app.get("/cases")
def get_cases():
with SessionLocal() as db:
rows = db.query(Case).order_by(Case.id.desc()).all()
return [
{"id": r.id, "created_at": r.created_at.isoformat(), "note": r.note}
for r in rows
]
@app.get("/cases/{case_id}")
def get_case(case_id: int):
with SessionLocal() as db:
c = db.get(Case, case_id)
if not c:
return {"error": "not found"}
containers = db.query(ContainerReport).filter_by(case_id=case_id).all()
images = db.query(ImageReport).filter_by(case_id=case_id).all()
findings = db.query(Finding).filter_by(case_id=case_id).all()
return {
"case": {"id": c.id, "created_at": c.created_at.isoformat(), "note": c.note},
"containers": [{
"id": r.id, "container_id": r.container_id, "name": r.name, "image": r.image,
"privileged": r.privileged, "runs_as_root": r.runs_as_root, "host_network": r.host_network,
"port_0_0_0_0": r.port_0_0_0_0, "added_caps": r.added_caps, "mounts": r.mounts,
"suspicious_env": r.suspicious_env
} for r in containers],
"images": [{
"id": r.id, "image_id": r.image_id, "repo_tags": r.repo_tags, "size_mb": r.size_mb,
"latest_tag": r.latest_tag, "age_days": r.age_days
} for r in images],
"findings": [{
"id": f.id, "subject_type": f.subject_type, "subject_id": f.subject_id,
"severity": f.severity, "title": f.title, "detail": f.detail
} for f in findings]
}
```
---
## Ruby dashboard (Sinatra)
### `ruby_dashboard/Gemfile`
```ruby
source 'https://rubygems.org'
gem 'sinatra'
gem 'thin'
gem 'sequel'
gem 'sqlite3'
```
### `ruby_dashboard/Dockerfile`
```dockerfile
FROM ruby:3.2
WORKDIR /app
COPY Gemfile ./
RUN bundle install
COPY . .
ENV RACK_ENV=production
ENV DB_PATH=/app/data/analyzer.db
CMD ["ruby", "app.rb", "-o", "0.0.0.0", "-p", "4567"]
```
### `ruby_dashboard/app.rb`
```ruby
require 'sinatra'
require 'sequel'
DB_PATH = ENV.fetch('DB_PATH', '/app/data/analyzer.db')
DB = Sequel.sqlite(DB_PATH)
get '/' do
@cases = DB[:cases].order(Sequel.desc(:id)).all
erb :index
end
get '/cases/:id' do
id = params[:id].to_i
@case = DB[:cases].where(id: id).first
halt 404, 'Not found' unless @case
@containers = DB[:container_reports].where(case_id: id).all
@images = DB[:image_reports].where(case_id: id).all
@findings = DB[:findings].where(case_id: id).all
erb :case
end
__END__
@@ layout
<!doctype html>
<html>
<head>
<meta charset="utf-8"/>
<title>Docker Cyber Analyzer</title>
<style>
body { font-family: system-ui, -apple-system, Segoe UI, Roboto, sans-serif; margin: 24px; }
header { display:flex; align-items:center; justify-content:space-between; }
.card { border:1px solid #ddd; border-radius:14px; padding:16px; margin-bottom:16px; box-shadow:0 1px 3px rgba(0,0,0,.05); }
code { background:#f6f8fa; padding:2px 6px; border-radius:6px; }
table { border-collapse:collapse; width:100%; }
th, td { border-bottom:1px solid #eee; padding:8px; text-align:left; }
.sev-LOW { color:#2c7; } .sev-MEDIUM { color:#fa0; } .sev-HIGH { color:#e55; } .sev-CRITICAL { color:#c00; font-weight:bold; }
</style>
</head>
<body>
<header>
<h1>Docker Cyber Analyzer</h1>
<a href="/">Home</a>
</header>
<%= yield %>
</body>
</html>
@@ index
<div class="card">
<h2>Cases</h2>
<ul>
<% @cases.each do |c| %>
<li>
<a href="/cases/<%= c[:id] %>">Case #<%= c[:id] %></a>
— <%= c[:created_at] %>
<% if c[:note] && !c[:note].empty? %> — <em><%= c[:note] %></em><% end %>
</li>
<% end %>
</ul>
<p>Trigger a scan via API: <code>curl -X POST http://localhost:8000/analyze</code></p>
</div>
@@ case
<div class="card">
<h2>Case #<%= @case[:id] %></h2>
<p><strong>Created:</strong> <%= @case[:created_at] %> — <strong>Note:</strong> <%= @case[:note] %></p>
</div>
<div class="card">
<h3>Findings</h3>
<table>
<thead><tr><th>Severity</th><th>Subject</th><th>Title</th><th>Detail</th></tr></thead>
<tbody>
<% @findings.each do |f| %>
<tr>
<td class="sev-<%= f[:severity] %>"><%= f[:severity] %></td>
<td><%= f[:subject_type] %> <code><%= f[:subject_id] %></code></td>
<td><%= f[:title] %></td>
<td><%= f[:detail] %></td>
</tr>
<% end %>
</tbody>
</table>
</div>
<div class="card">
<h3>Containers</h3>
<table>
<thead>
<tr><th>ID</th><th>Name</th><th>Image</th><th>Privileged</th><th>Root</th><th>HostNet</th><th>PublicPorts</th><th>Caps</th><th>Mounts</th><th>Suspicious Env</th></tr>
</thead>
<tbody>
<% @containers.each do |c| %>
<tr>
<td><code><%= c[:container_id] %></code></td>
<td><%= c[:name] %></td>
<td><%= c[:image] %></td>
<td><%= c[:privileged] %></td>
<td><%= c[:runs_as_root] %></td>
<td><%= c[:host_network] %></td>
<td><%= c[:port_0_0_0_0] %></td>
<td><%= c[:added_caps] %></td>
<td><%= c[:mounts] %></td>
<td><%= c[:suspicious_env] %></td>
</tr>
<% end %>
</tbody>
</table>
</div>
<div class="card">
<h3>Images</h3>
<table>
<thead><tr><th>ID</th><th>Repo Tags</th><th>Size (MB)</th><th>Latest?</th><th>Age (days)</th></tr></thead>
<tbody>
<% @images.each do |i| %>
<tr>
<td><code><%= i[:image_id] %></code></td>
<td><%= i[:repo_tags] %></td>
<td><%= i[:size_mb] %></td>
<td><%= i[:latest_tag] %></td>
<td><%= i[:age_days] %></td>
</tr>
<% end %>
</tbody>
</table>
</div>
```
---
## Orchestrator
### `docker-compose.yml`
```yaml
services:
python_api:
build: ./python_service
ports:
- "8000:8000"
volumes:
- /var/run/docker.sock:/var/run/docker.sock
- ./data:/app/data
environment:
- DB_PATH=/app/data/analyzer.db
- ANALYZER_IMAGE_AGE_DAYS=365
- ANALYZER_IMAGE_SIZE_MB=800
ruby_ui:
build: ./ruby_dashboard
ports:
- "4567:4567"
volumes:
- ./data:/app/data
depends_on:
- python_api
```
---
## Extending with AI/ML (ideas)
* Create a `features` table and export normalized features per container/image.
* Train a classifier (e.g., XGBoost) to label HIGH/MEDIUM/LOW risk using your historical cases.
* Add an `/score` endpoint that returns a composite risk score and recommended actions.
* Add YARA/Trivy/Grype adapters for malware/CVE signals and feed their outputs as features.
---
## Safety & Ops notes
* Mounting `/var/run/docker.sock` gives the analyzer broad control—limit to lab/test machines.
* Prefer read-only host mounts and least-privilege where possible.
* Consider Postgres for multi-user deployments; swap the SQLAlchemy URL and update the Ruby Sequel URL accordingly.
Reference in New Issue View Git Blame Copy Permalink