tool--656ed9e6-0994-563e-eed8-410639322500 · TLP:CLEAR
CoverageIQ
Full-stack security platform concept that predicts likely next attacker moves from MITRE ATT&CK campaign data, then checks Sigma detection coverage for each predicted technique.
Working demo; the alerts are simulated
Overview
What it does
Every SOC has the same gap: an alert fires on a phishing technique, and you're left guessing what comes next. Your SIEM caught the entry point — but if the attacker pivots to credential dumping or lateral movement, do you have rules for that? Most teams find out the answer after the breach.
CoverageIQ answers two questions on every alert:
- What technique is likely coming next? (based on the 50+ campaigns and 170+ groups documented in MITRE ATT&CK, not guesswork)
- Do I actually have Sigma detection coverage for it?
Illustrative output, not measured results:
Alert fires → T1566 (Phishing detected)
↓
Predict next moves:
→ T1055 Process Injection 72% ⚠️ NO DETECTION RULE
→ T1078 Valid Accounts 61% ✅ 2 Sigma rules
→ T1003 Credential Dumping 44% ⚠️ NO DETECTION RULE
↓
"You are blind to 2 of 3 most likely next moves"
Key features
Attack Chain Predictor — When an alert fires, the detail page shows the 3 most likely follow-on techniques with probability scores derived from historical ATT&CK campaign sequences.
Detection Coverage Heatmap — Interactive ATT&CK matrix showing every technique colored by your Sigma rule coverage. Click any cell to see which rules cover it, which actors use it, and whether Atomic Red Team tests exist.
Threat Actor Coverage Scoring — Pick any APT group and instantly see how well your current rule set would detect their full TTP profile.
PDF Threat Intel Ingestion — Upload CISA advisories or APT reports. The platform extracts ATT&CK technique IDs (T#### pattern), matches them against your Sigma library, and returns a gap report in seconds.
Alert Simulator — Built-in simulator generates realistic alerts every 30 minutes for demo and testing without needing live log sources.
Architecture
The full stack runs in Docker Compose — six services, no external paid dependencies:
- FastAPI backend — prediction engine, coverage calculation, PDF ingestion
- PostgreSQL — techniques, campaigns, coverage scores, alerts
- Redis — task queue and caching
- Celery worker + Beat — async data sync and scheduled coverage recalculation
- Next.js frontend — ATT&CK heatmap, alert queue, actor coverage dashboard
All data sources are free: MITRE ATT&CK STIX data published by MITRE, SigmaHQ rules from GitHub, Atomic Red Team from GitHub.
What I learned
Building this exposed how large the gap is between detection rule libraries and real attacker behavior. Teams have thousands of Sigma rules but no systematic way to check whether those rules cover what attackers actually do next. The prediction model built on real campaign data is a more honest signal than any heuristic playbook.
Objectives
- Predict attacker next moves from real MITRE ATT&CK campaign sequence data
- Map predicted techniques against live Sigma rule library to surface detection blind spots
- Ingest CISA advisories and APT reports via PDF upload for instant coverage analysis
- Score threat actor coverage — 'how well would you detect APT29 right now?'
- Build a deployable full-stack security platform with realistic data flows and analyst-facing workflows
Tools Used
Methodology
- Ingested the MITRE ATT&CK Enterprise STIX bundle (50+ campaigns, 170+ groups, 700+ techniques and sub-techniques)
- Built technique chain model from campaign sequence data to power next-move prediction
- Loaded and indexed 3,000+ SigmaHQ rules by ATT&CK technique tag
- Built PDF ingestion pipeline using regex T#### pattern extraction for CISA advisories
- Deployed full stack via Docker Compose: Postgres, Redis, Celery worker, Celery Beat, FastAPI, Next.js
- Implemented an alert simulator for demo and testing workflows
MITRE ATT&CK Mapping
| Technique | Tactic | Description |
|---|---|---|
| T1566 - Phishing | Initial Access | Example trigger technique — platform predicts T1055, T1078, T1003 as next moves |
| T1055 - Process Injection | Defense Evasion, Privilege Escalation | Example of a predicted follow-on to phishing with a coverage gap |
| T1003 - OS Credential Dumping | Credential Access | Example of a predicted follow-on with Sigma rule gaps flagged |
| T1078 - Valid Accounts | Defense Evasion, Persistence | Example of a predicted follow-on where existing Sigma rules were found |
Key Findings
- Alert triage gap: most SOC teams respond to the fired technique but have no coverage model for follow-on moves
- Real MITRE ATT&CK campaign data gives statistically grounded next-move predictions, not heuristic guesses
- PDF ingestion (CISA advisories, APT reports) enables rapid coverage analysis without manual TTP extraction
- Sigma coverage for commonly predicted follow-on techniques is uneven: some have several rules, others none