Skip to content

r0ms3c/securepulse

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SecurePulse

AI-powered security intelligence platform that ingests CVEs, vendor advisories, and threat intel feeds, scores them by severity and exploitability using LLMs, and delivers a prioritized daily briefing via email, Slack, and a web dashboard.

"Wake up to a security briefing that's already been triaged. Read only what matters. Miss nothing critical."


The Problem

Security engineers start every day drowning in noise — CVE databases, vendor advisories, threat intel blogs, CISA alerts. By the time you've manually triaged what matters, you've lost 1-2 hours. And you've probably still missed something.

The Solution

SecurePulse runs continuously in the background, ingesting from multiple sources, scoring each item across five risk dimensions using an LLM, and delivering a prioritized briefing every morning. It acts as your first-pass security analyst — one that never sleeps and never skips a source.


Architecture

EventBridge (cron)
        │
        ▼
Fetcher Lambdas (NVD, CISA KEV, EPSS, RSS)
        │
        ▼
SQS Queue
        │
        ▼
Scorer + LLM Enricher Lambdas
        │
        ▼
DynamoDB + S3
        │
        ▼
Email (SES) + Slack + Web Dashboard

Built entirely on AWS Serverless — Lambda, DynamoDB, SQS, EventBridge, SES, S3, CloudFront. Infrastructure as Code with AWS CDK (Python).


Scoring Framework

Each item is scored across five dimensions:

Dimension Weight Signal
Exploitability 30% EPSS score, CISA KEV status
Severity 25% CVSS base score, CWE type
Freshness 15% Publication date, patch availability
Breadth 15% Affected products, vendor market share
Intel Value 15% IOCs, TTPs, mitigations present

Priority tiers: CRITICAL (8.0-10.0) → HIGH (6.0-7.9) → MEDIUM (4.0-5.9) → LOW (1.0-3.9)


Feed Sources

Phase 1 (active): NVD/CVE, CISA KEV, EPSS Phase 2: Microsoft MSRC, AWS Security, Google Project Zero, vendor RSS Phase 3: Mandiant, CrowdStrike, Unit 42, Krebs on Security Phase 4: Commercial threat intel APIs (Recorded Future, AlienVault OTX)


Project Status

Phase Description Status
1 Foundation — NVD + CISA KEV + EPSS → email digest 🔨 In Progress
2 Enrichment — vendor advisories + Slack delivery ⏳ Planned
3 Dashboard — React web UI with search and filters ⏳ Planned
4 Advanced Intel — threat blogs + IOC extraction ⏳ Planned
5 Productization — multi-tenancy + billing ⏳ Planned
6 OpenClaw — conversational AI agent layer ⏳ Planned
7 Portfolio — showcase, blog post, demo video ⏳ Planned

Cost

Approximately $10-25/month for a solo user on AWS (eu-west-1). LLM API calls are the dominant cost — using AWS Bedrock reduces this significantly.


Tech Stack

  • Infrastructure: AWS CDK (Python), CloudFormation
  • Compute: AWS Lambda (Python 3.12)
  • Queue: AWS SQS with Dead Letter Queue
  • Database: AWS DynamoDB (on-demand)
  • Storage: AWS S3
  • Email: AWS SES
  • Scheduling: AWS EventBridge
  • AI/LLM: Provider-agnostic gateway (Claude, GPT-4, AWS Bedrock)
  • Dashboard: React + Vite + Tailwind (Phase 3)

Setup

Full setup instructions coming in Issue #13 once Phase 1 is complete. Prerequisites: Python 3.12+, Node.js 22+, AWS CLI v2, AWS CDK CLI.


License

MIT

About

AI-powered security intelligence platform that ingests CVEs, vendor advisories, and threat intel feeds, scores them by severity and exploitability using LLMs, and delivers a prioritized daily briefing via email, Slack, and a web dashboard. Built on AWS Serverless (Lambda, DynamoDB, SQS, CDK).

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors