KoolAdeOps 🥤

Automated research pipeline that pulls cybersecurity articles from 19 curated sources, scores them for relevance, summarizes them with a local LLM through a cybersecurity analyst lens, and delivers a daily Markdown briefing — no subscriptions, no cloud API keys required.

Keeping Up Is a Full-Time Job

The AI security and DevSecOps landscape moves fast. ArXiv drops new papers daily. CISA publishes advisories. Schneier, Krebs, The Hacker News, Reddit communities, and vendor blogs all publish independently. Manually checking each source, filtering signal from noise, and distilling what actually matters takes hours — time most engineers don't have.

What KoolAdeOps Does

KoolAdeOps runs as a single command. It fetches RSS feeds from 19 sources across academic research, security advisories, community discussion, and vendor blogs. Each article is scored against a curated keyword list covering prompt injection, SBOM, CI/CD pipeline security, LLM vulnerabilities, container security, and more. High-signal articles are stored in a local SQLite database, optionally summarized using a locally running Ollama model (gemma4 by default) through a cybersecurity analyst lens — focusing on vulnerabilities, threat actors, affected systems, severity, and recommended mitigations. Non-cybersecurity articles are automatically filtered out at the summarization stage. The briefing includes top highlights with relevance scores, articles grouped by category, a 30-day trend tracker, and a source breakdown table. Everything runs locally — no data leaves your machine.

Sample Output

Running python main.py --days 3 produces a file like briefings/briefing-2026-05-17.md:

# AI & DevSecOps Research Briefing — 2026-05-17

*Generated 2026-05-17 08:14 UTC | 42 relevant articles from the last 3 days*

## Top Highlights

### [Prompt Injection Attacks Against Code Assistants](https://arxiv.org/abs/...)
**Source:** ArXiv CS.CR | **Relevance:** 0.74

Researchers found that popular AI code assistants can be manipulated through
malicious comments in dependencies to generate insecure code. Affects any
developer using AI-assisted coding tools against untrusted repositories.
Mitigation: review AI suggestions against source context before accepting.

**Topics:** `prompt injection`, `ai code generation security`, `supply chain`

---

## Trending Topics (30-day window)

**Sustained topics** (5+ occurrences):
- `llm security` — seen 14 times
- `software supply chain` — seen 9 times

**Rising topics** (2-4 occurrences):
- `ai soc` — seen 3 times

Usage

Prerequisites: Python 3.10+, and Ollama running locally if you want LLM summaries.

git clone https://github.com/tdiprima/KoolAdeOps.git
cd KoolAdeOps
pip install -r requirements.txt

Run the full pipeline (fetch → score → summarize → briefing):

python main.py

Skip LLM summarization (useful if Ollama isn't running):

python main.py --no-llm

Regenerate today's briefing from existing data (no network calls):

python main.py --briefing

Set the lookback window (default is 7 days):

python main.py --days 3

Export articles for a specific day to a Markdown file:

python export_articles.py

The script lists the dates available in the database, then prompts you to pick one:

Available dates in database:
  2026-05-17
  2026-05-16
  2026-05-15

Enter a date to export (YYYY-MM-DD), 'today' (2026-05-17), or 'yesterday' (2026-05-16).
Date [today]:

Press Enter to accept today, type yesterday, or enter any YYYY-MM-DD date. The output is written to articles_export_YYYY-MM-DD.md in the project root.

Configuration via environment variables:

Variable	Default	Description
OLLAMA_MODEL	gemma4	Local model name
OLLAMA_BASE_URL	http://localhost:11434	Ollama endpoint (localhost only)
MAX_ARTICLE_AGE_DAYS	7	Age cutoff for fetched articles
MAX_ARTICLES_PER_FEED	20	Cap per RSS source
FETCH_TIMEOUT	30	HTTP timeout in seconds

Briefings are written to the briefings/ directory, named briefing-YYYY-MM-DD.md. The SQLite database (research.db) accumulates over time, enabling the 30-day trend analysis to improve with each run.

Disclaimer

"KoolAdeOps" is an independent open-source project and is not affiliated with, endorsed by, sponsored by, or connected to Kool-Aid, Kraft Heinz, or any of their subsidiaries or brands.

This repository exists purely for educational purposes. No powdered beverages were harmed, mixed, or distributed in the making of this project. 👻