A cybersecurity tools aggregator inspired by KitPloit. Automatically discovers security tools from GitHub, generates AI summaries, and publishes a static site via Hugo + GitHub Pages.

Fork this repo and edit hackyfeed.toml to create your own tool aggregator for any topic.

GitHub API ──→ SQLite DB ──→ AI Summaries ──→ Hugo Markdown ──→ GitHub Pages
  (topics)      (diff)       (LiteLLM)        (generate)        (deploy)

A single Go CLI tool handles the entire pipeline:

• hackyfeed fetch — Searches GitHub for repos by topic and parses awesome lists. Upserts into SQLite.
• hackyfeed summarize — Fetches READMEs and generates summaries via any OpenAI-compatible API.
• hackyfeed generate — Renders Hugo markdown from the database.
• hackyfeed all — Runs fetch → summarize → generate in sequence.

• Go 1.22+
• Hugo (extended)
• A GitHub personal access token
• An OpenAI-compatible API endpoint (LiteLLM, OpenAI, Ollama, etc.)

git clone https://github.com/rainmana/hackyfeed.git
cd hackyfeed
cp .env.example .env
# Edit .env with your tokens

go build -o hackyfeed ./cmd/hackyfeed/
export $(grep -v '^#' .env | xargs)
./hackyfeed all
cd site && hugo server

go test ./... -v

The entire pipeline is driven by hackyfeed.toml. Fork this repo and edit it to aggregate anything:

[fetch]
topics = [
  "machine-learning",
  "deep-learning",
  "llm",
  "transformer",
]
min_stars = 50

[fetch]
awesome_lists = [
  "https://raw.githubusercontent.com/yourname/awesome-list/main/README.md",
]

[categories]
default_category = "ai-tools"

[categories.rules]
nlp = ["nlp", "natural-language", "text-processing"]
vision = ["computer-vision", "image", "object-detection"]
llm = ["llm", "large-language-model", "gpt", "transformer"]

[summarize]
system_prompt = """You are an AI/ML tools cataloger. Given a README, produce JSON with:
- "summary": 2-3 sentence description for a technical audience.
- "install": Primary installation method.
Respond ONLY with valid JSON."""
max_readme_chars = 4000

[site]
title = "ML Feed"
description = "Discover the latest machine learning tools from GitHub."
author = "yourname"

Then update site/hugo.toml to match your site title/URL.

The included workflow runs daily at 8am MT (2pm UTC). Set these repo secrets:

• GH_PAT — GitHub personal access token
• LLM_API_BASE — Your LLM API endpoint
• LLM_API_KEY — Your LLM API key

In repo Settings → Pages, set source to GitHub Actions.

├── cmd/hackyfeed/       # CLI entry point
├── internal/
│   ├── config/          # TOML config loader
│   ├── db/              # SQLite schema & queries
│   ├── fetch/           # GitHub API + awesome list parser
│   ├── summarize/       # LLM-powered README summarization
│   └── generate/        # Hugo markdown generator
├── site/                # Hugo site
│   ├── content/         # Generated tool pages
│   └── themes/          # Terminal hacker theme (dark/light)
├── hackyfeed.toml       # ← Edit this to customize everything
├── .github/workflows/   # Daily automation
└── .env.example         # Environment variable template

HackyFeed is an aggregator — inclusion is not endorsement. All tools must be used ethically and legally. Nothing on this site represents the views of the maintainer's employer. See the full disclaimer.

MIT