Welcome to The Weekly Five - your curated list of 5
exceptional open source projects I discovered this week.
The Weekly Five
Take back control: tools that let you own your stack, your data, and your workflow
The promise of modern software often comes with a hidden cost. Your data ends up scattered across services you don't control, workflows get locked into platforms that can change terms overnight, and systems pile up with apps you never asked for. This week's picks flip that script. From stripping Windows down to essentials, to building your own document archive and dashboard, to harnessing AI agents and bulletproof data validation, these five tools share a common thread: they put you back in the driver's seat.
Top takeaways
Owning your stack doesn't require deep technical expertise (several of these tools prioritize drag-and-drop interfaces, one-command installs, and no-YAML configuration)
Self-hosted solutions have matured significantly, with features like OCR, email ingestion, and 40+ integrations now available out of the box
Data validation and AI workflow tooling are converging, making it easier than ever to build reliable, automated pipelines you fully control
Who this issue is for
Developers, homelabbers, and privacy-conscious power users who want practical tools to reclaim ownership of their digital environment without sacrificing convenience.
Win11Debloat
Why this made the cut: With nearly 50K stars, this is the definitive community-driven solution for stripping Windows of unwanted bloat, telemetry, and clutter.
Why it matters
A fresh Windows installation comes loaded with pre-installed apps, background telemetry, and advertising hooks that erode both performance and privacy. Win11Debloat gives you a single, lightweight PowerShell script to reclaim your operating system without needing to reinstall or manually hunt through settings menus.
Key features
Removes pre-installed bloatware (Candy Crush, Xbox apps, news widgets, and more)
Disables telemetry and data collection with registry-level tweaks
Works on both Windows 10 and Windows 11 with no installation required
Offers both automated and interactive modes for granular control
Community-maintained with regular updates for new Windows builds
How to use
Download the latest release from GitHub or clone the repository
Right-click the script and select "Run with PowerShell" (or run from an elevated terminal)
Choose interactive mode to review each change, or automated mode for a one-click cleanup
Create a system restore point before running, then review the script's options to skip any apps you actually use (this step saves headaches if you accidentally remove something you need)
The interactive mode walks you through each category of changes: start there if you're unsure what to remove, then switch to automated mode for future reinstalls once you know your preferences
🔗 View on GitHub | GitHub stars: 49,803
📖 Resources: Wiki & how-to guides | Command-line interface reference | Reverting changes
papra
Why this made the cut: Papra brings enterprise-grade document management (OCR, email ingestion, real API) to a minimalist, self-hosted package anyone can run.
Why it matters
Scattered PDFs, receipts, and contracts across cloud drives and email threads mean you never truly own your documents. Papra centralizes everything in a self-hosted archive you control, with powerful features like OCR search and email ingestion that rival commercial solutions.
Key features
Self-hosted document archiving with a clean, minimalist interface
Built-in OCR for searching scanned documents and images
Email ingestion to automatically capture attachments
Real API for integrating with other tools and automations
Lightweight TypeScript codebase, easy to deploy via Docker
How to use
Deploy Papra using Docker (compose file available in the repo)
Configure email ingestion by connecting your preferred inbox
Upload documents manually or let them flow in via email
Use the search bar to query text inside scanned files (OCR happens automatically on upload)
Organize documents into folders early to keep your archive manageable as it grows. This becomes critical once you have hundreds of files flowing in via email
The API opens up powerful automations: connect it to your receipt scanner app or expense workflow to eliminate manual uploads entirely
🔗 View on GitHub | GitHub stars: 4,869
📖 Resources: Self-hosting docs | Email ingestion guide | Live demo (no backend)
homarr
Why this made the cut: Homarr makes self-hosted dashboards accessible to everyone with drag-and-drop configuration, 40+ integrations, and no YAML required.
Why it matters
Running multiple self-hosted services quickly turns into tab chaos. Homarr unifies your homelab into a single, beautiful dashboard with built-in authentication, real-time service status, and deep integrations with tools like Sonarr, Radarr, and Plex. All without writing a single line of YAML.
Key features
Drag-and-drop dashboard builder with 11K+ built-in icons
40+ integrations for popular self-hosted apps (Docker, Servarr suite, media servers)
Authentication out of the box for secure access
Privacy-focused: runs entirely on your hardware
Active development with 74 contributors and growing
How to use
Deploy Homarr via Docker (official image available on Docker Hub)
Access the web interface and start adding widgets by dragging them onto your dashboard
Connect integrations by entering service URLs and API keys (no config files to edit)
Customize themes, layouts, and icons to match your preferences
Group related services (media, networking, utilities) into sections for faster navigation. This organizational step pays off immediately when your homelab grows beyond a handful of services
The built-in icon library covers most popular self-hosted apps, so you rarely need to hunt for custom icons
🔗 View on GitHub | GitHub stars: 4,200
📖 Resources: Installation docs | Homarr 1.0 announcement | Live demo
Trellis
Why this made the cut: As AI agents become central to developer workflows, Trellis offers a structured engineering framework that keeps you in control of how those agents operate.
Why it matters
AI coding assistants like Claude Code and Codex are transforming development, but each new session starts from scratch with no memory of your project, your conventions, or your team's requirements. Trellis persists specs, tasks, and memory into your repo, so any coding agent works to your engineering standards instead of winging it.
Key features
Engineering framework that injects project specs, tasks, and memory into every AI coding session
Supports 16 AI coding platforms including Claude Code, Codex, Cursor, and Windsurf
Structured 4-phase workflow: Plan, Implement, Verify, and Finish
TypeScript-based for easy customization and extension
Growing community (37 contributors, 11K+ stars)
How to use
Install Trellis globally via npm: npm install -g @mindfoldhq/trellis@latest
Initialize in your repo with trellis init -u your-name (add platform flags like --cursor or --codex for the tools you use)
Describe what you want in natural language and let Trellis brainstorm requirements with you before implementation begins
The AI runs through Plan, Implement, and Verify phases automatically, checking results against your specs, lint, type-check, and tests
Start with a simple single-task workflow before scaling to complex multi-step projects. The examples folder in the repo contains starter templates that demonstrate common patterns
Because specs live in .trellis/ inside your repo, the same structure works across all supported platforms, making it easy to test different AI tools on the same task
🔗 View on GitHub | GitHub stars: 11,366
pydantic
Why this made the cut: Pydantic is the gold standard for Python data validation, and owning your data pipeline starts with knowing your data is clean.
Why it matters
Garbage in, garbage out. Whether you're building APIs, processing user input, or feeding data into AI models, unvalidated data introduces bugs, security holes, and unpredictable behavior. Pydantic uses Python type hints to enforce data contracts at runtime, catching errors before they cascade.
Key features
Data validation powered by Python type hints (no separate schema language)
Automatic JSON Schema generation for API documentation and interoperability
Deep integration with FastAPI (the most popular Python web framework for APIs)
Supports Python 3.10 through 3.13
Battle-tested with 761 contributors and 28K+ stars
How to use
Install pydantic via pip: pip install pydantic
Define a model by subclassing BaseModel and adding typed fields:
from pydantic import BaseModel
class User(BaseModel):
name: str
email: str
age: int
Instantiate the model with data: invalid input raises clear, actionable errors
Use .model_dump() to serialize validated data to dictionaries or JSON
Use Field() to add constraints like min/max length, regex patterns, and default values directly in your model definitions. This keeps validation logic co-located with your data structure instead of scattered across your codebase
Nested models work naturally (just use one model as a type hint inside another), which makes validating complex API payloads straightforward
🔗 View on GitHub | GitHub stars: 28,149
📖 Resources: Getting started | Why use Pydantic | Pydantic blog
If you only try one
Start with Win11Debloat. It requires no ongoing maintenance, takes just minutes to run, and the payoff is immediate: a cleaner, faster, more private Windows experience. Unlike tools that demand infrastructure or ongoing configuration, Win11Debloat is a one-and-done improvement that benefits every session you spend on your machine. Once you've reclaimed your OS, you'll have the momentum (and the cleaner system) to tackle the rest of your stack.
If you are doing Open Source I have a good news for you, I work at CodeRabbit which is an AI review tool and its free for Open Source, please reach out to me on X or LinkedIn or just send an email on [email protected] if you need help on adopting CodeRabbit.
You can visit our portal below to create a new account and connect your repository and start reviewing your code.
