Hands-on SOC lab: AWS mini-SOC with CloudTrail/GuardDuty, a Python CloudTrail IOC analyzer, phishing header forensics, ELF binary triage, and agent skills.
👉 New here? Read START_HERE.md first. It explains what this repo is, gives you a path to follow (new to SOC / practicing detection / prepping for interviews), and has one table of every case with its CVE/KEV status, skills, contents, and the command to verify its detection.
Hands-on cybersecurity lab projects: a mini-SOC on AWS, a Python IOC analyzer, a phishing analysis lab, binary triage, agent skills, a VS Code supply-chain defense lab, KEV-driven threat-hunting packs, and a Node-RED misconfiguration flow-audit lab. Built to demonstrate verifiable, practical security skills.
🔗 https://github.com/szzxli/soc-home-lab
No AWS account needed to start. Pick the path that matches you — each is a short sequence, not the whole repo.
Path 1 — Start here, no AWS needed (browser only):
Path 2 — Practicing detection (Python, local only):
git clone https://github.com/szzxli/soc-home-lab.git
cd soc-home-lab/02-python-ioc-analyzer
pip install -r requirements.txt
python ioc_analyzer.py --input sample_data/sample_cloudtrail.json --output report.md
Then pick a noisy KEV pack and write your own detection before comparing with the reference script.
Path 3 — AWS mini-SOC (optional):
Lab 01: mini-soc is the only lab that needs your own AWS account, and it costs money — see Cost & cleanup. Optional: do it after Paths 1–2, or skip it.
Browse the full lab site: Project site
📊 Skills quantified: these projects are graded against a public rubric and mapped to the NIST NICE framework in the companion repo szzxli/soc-skills — including a one-page HR-readable skill profile.
All datasets, log samples, and email artifacts in this repo are synthetic training samples — clearly labeled as such. No real incidents, real logs, or real phishing emails are presented here.
What this is / isn't: beginner-to-intermediate SOC practice on synthetic data (clean samples plus noisy variants with honest TP/FP numbers). It isn't proof of job readiness and isn't real SOC volume.
What's different here: KEV-driven, with every CVE / KEV claim re-verified weekly against the official CISA feed (tools/verify_kev.py, workflow) · the demo runs in your browser · pairs with the soc-skills gap analysis.
Problem: SOC analysts need to know how cloud detections actually fire — not just theory. What it is: A minimal AWS detection stack (VPC → CloudTrail → S3 → GuardDuty → SecurityHub), three adversary-emulation scenarios using Stratus Red Team (verified technique IDs), a triage guide, and blank incident report templates. Skills: CloudTrail analysis, GuardDuty triage, incident response workflow, Terraform IaC.
Problem: Tier-1 analysts drown in raw CloudTrail logs; simple heuristics catch the obvious badness fast. What it is: A working CLI tool that parses CloudTrail JSON, flags anomalous logins / failed-login spikes / risky IAM changes, enriches IPs via AbuseIPDB, and emits a Markdown report. Ships with synthetic sample data and passing pytest suite. Skills: Python scripting, log analysis, IOC enrichment, detection heuristics.
Problem: Phishing triage is daily Tier-1 SOC work — header forensics, auth-result interpretation, URL deobfuscation. What it is: A full walkthrough of a reconstructed training-sample phish, a header-analysis cheatsheet, a blank analysis report template, and a self-test guide using GoPhish (own mailbox only). Skills: Email header forensics, SPF/DKIM/DMARC, URL analysis, phishing triage.
Problem: Agent-facing reverse-engineering CLIs expose dozens of granular subcommands and machine-oriented JSON — powerful, but not human-friendly for quick triage.
What it is: A one-command ELF binary triage CLI (triage.py <binary>): section-entropy packing heuristics, W^X detection, suspicious-import watchlist, string IOC extraction, score-based verdict (benign/suspicious/malicious), and a one-page analyst report. Optional third-party inspect-binary-layout enrichment when a provider is configured; degrades honestly without it.
Skills: Static binary analysis, ELF internals, malware triage heuristics, detection engineering.
Problem: General-purpose agent skill packs don't cover SOC work.
What it is: Seven agent skills for the Tier-1 analyst loop, distilled from working SOC practice in this repo: alert-triage, incident-report, cloudtrail-ioc, phishing-analysis, binary-triage, detection-engineering, yaral-detection. Each skill declares its trigger, workflow, output contract, and honesty boundaries. MIT licensed, copy-paste installable into any agent's skills directory.
Skills: Detection triage, incident documentation, knowledge distillation into agent skills.
Problem: Supply-chain attacks now target the developer's own workbench — a malicious .vscode/tasks.json can execute code the moment a project folder is opened, and a cron-based backdoor survives deleting the repo files.
What it is: A safe, hands-on supply-chain defense lab built around the GhostAction technique (October 2026): inert training samples that demonstrate the mechanism with zero malicious behavior, a hunt checklist for the two persistence points (workspace tasks + cron), four exercises, and a blank incident report template.
Skills: Supply-chain threat modeling, static review of editor configs, persistence hunting (cron/systemd), incident documentation.