6 open-source gems I keep to myself: server monitoring, PDF, Markdown and PPT, bundled in one post

Title options (pick 1 of 3)

6 open-source gems I keep to myself: server monitoring, PDF, Markdown and PPT, bundled in one post

From missed renewals to painful slide decks: 6 open-source projects that fix both in one go

These 6 GitHub projects save me two hours of overtime every day

1. Intro: from pain to payoff, one project away

After a few years in tech, the number of "expiry dates" I have to track keeps growing. Domains need renewing, SSL certificates need swapping, each VPS bills on its own cycle, and the team's SaaS subscriptions are a mess. Every month one or two services get suspended because I forgot to renew them. I stare at a wall of red monitoring alerts and quietly pay up. Then I swear I'll remember next time — and forget again the time after.

What I was really missing wasn't memory. It was a tool that pulls all those expiry dates — scattered across admin panels, inboxes and text messages — into one calendar view and pushes them in front of me. Over the past year I dug out the open-source projects I'd been hoarding and tried each of them. Here are the 6 most worth sharing, in the hope that you skip a few of the potholes I hit.

2. The shortlist

3. Project breakdowns

ieax/renewhelper — an expiry-date keeper that lives on Cloudflare

RenewHelper solves exactly the pain point from the intro. It puts every recurring subscription, domain, server and certificate expiry date into a single panel deployed on Cloudflare Workers, works out lunar/solar recurrence cycles automatically, and pushes reminders through a dozen-plus channels including Telegram, Bark, DingTalk, Feishu and email. It also generates an .ics subscription link you can sync to iOS/Google Calendar. Since v2.x it ships a cash-flow dashboard, tracking billed amounts separately from actual spending, with automatic multi-currency conversion. The frontend and backend are fully decoupled, it runs standalone without any CDN, and you can also deploy it locally with Docker.

The technical highlight is using Cloudflare Workers + KV, a pure serverless stack, to hold state. The free tier is enough for personal use, so no VPS purchase is needed. A Cron trigger runs on 0,30 * * * * to scan expiring items every half hour, then auto-renews or marks them disabled.

Who it's for:

• Individual developers juggling more than 5 domains, VPS instances or SaaS subscriptions
• People who want to ride Cloudflare's free tier instead of writing their own cron jobs

Who it's not for:

• Team-level asset and contract management (the data model is personal-oriented)
• Enterprise scenarios that need complex approval flows and role permissions

Quickest way in

One-click deploy is recommended: hit the Deploy to Cloudflare Workers button in the README and authorize GitHub. ==Note: you must bind the KV namespace to the variable name RENEW_KV in the Worker settings, otherwise nothing gets saved.==

Once deployed, visit the Worker domain. The default password is admin — change it right after you log in.

0,30 * * * *

Drop that Cron expression into Triggers → Cron Triggers and the scheduled scan goes live.

Source: https://github.com/ieax/renewhelper

huilang-me/CF-Server-Monitor — a free multi-server monitoring panel

If you want to know whether each server is still alive, want historical CPU/memory/network curves, and don't want to spin up a VPS just for that, CF-Server-Monitor is built for you. It stuffs the dashboard, API, database and real-time push into Cloudflare: data lands in D1, real-time push goes through Durable Objects + WebSocket, and the frontend is Vue. On the server side you install only a lightweight agent (written in Go, reporting every 60 seconds by default). Reporting is one-way — no WebSSH, no remote command execution — so the security model is noticeably cleaner than a classic controller/controlled probe setup.

The default 60-second interval handles roughly 60 machines; switch to 120 seconds and you can double that. The agent supports Linux, Alpine, OpenWrt, Synology DSM, fnOS, macOS and Windows, and supports non-root installs (systemd --user), so a regular user account is enough.

Who it's for:

• People with several VPS instances or home servers who want one panel showing overall health
• Users who care about security and privacy and don't want a "controller + controlled" architecture

Who it's not for:

• Ops teams that need process-level monitoring, log aggregation or APM
• People who want a ready-made SaaS and refuse to install an agent on every machine

Quickest way in

Connecting the repo to Cloudflare Workers gets you there in one step: log into the Cloudflare dashboard → Workers & Pages → Create application → Create Worker, paste the code in, create the D1 database and fill in API_SECRET.

On the server side, bring the agent up with the official install command:

curl -sSL https://raw.githubusercontent.com/huilang-me/cfsm-agent/main/install.sh | bash -s -- --server https://your-worker.workers.dev --secret YOUR_API_SECRET

==Note: the agent validates API_SECRET against the Worker at install time, so the secret must match the backend.==

Source: https://github.com/huilang-me/CF-Server-Monitor

nyblnet/bento — an entire office suite inside one HTML file

Bento takes the opposite approach to everything above: it isn't deployed on any server, because the file itself is the software. Open a .bento.html file and it contains the viewer, the presenter and the editor — double-click to edit, hit F5 to present, email it as an attachment and the recipient needs nothing installed. The whole file is about 560 KB, with the renderer, animation engine, chart engine and collaborative CRDT all packed inside.

At the top of the file sits a plain-text JSON block. The data is JSON, with no binary format anywhere. That "view-source transparent" stance buys two things. First, an AI agent can read and write that JSON directly to edit slides (the project ships a Claude Code skill). Second, offline-first: with Offline mode on, sync, collaboration and downloads are hard-blocked, so data never leaves your machine.

Who it's for:

• People fed up with PPT/Keynote compatibility who want "one file + any browser" presentations
• Users who'd rather hand slides to an AI agent for revisions than tweak styles by hand
• Anyone with strong opinions about local-first data and readable source

Who it's not for:

• Traditional office setups that depend on Office 365 co-editing and version history
• Scenarios that need complex VBA macros or third-party Office plugins

Quickest way in

Download Bento_Slides.bento.html from the Releases page and double-click it to edit in the browser.

After installing the Claude Code skill, you can have the agent revise slides in one shot:

/plugin marketplace add nyblnet/bento

Then tell the agent what you need:

Replace the pie chart on this slide with a bar chart and keep the data

Source: https://github.com/nyblnet/bento

alexishida/Moji — opens Markdown the way you open a PDF

Most people who write documents in Markdown have hit the same wall: you send the file over and the other person has no editor installed, so all they see is raw syntax. Moji is an Electron desktop app whose whole goal is "double-click a .md file and it opens, WYSIWYG". It's built on Electron + React + TypeScript and includes an editor (CodeMirror 6), preview, outline navigation, search and replace, HTML/PDF/PNG export, and full Mermaid flowchart/sequence diagram rendering (7 diagram types, zoomable, draggable and exportable to PNG after rendering).

There's real engineering effort around large documents: the main process reads in streamed chunks, Markdown is parsed inside a Web Worker, and files over 20 MB automatically switch to block-level virtualization. Multiple tabs, dirty markers and unsaved-change prompts are all there too, so it isn't "yet another toy Markdown editor".

Who it's for:

• People who write Markdown docs, notes and technical blogs every day
• Users who need to export .md to PDF/HTML in one click and send it to non-technical colleagues

Who it's not for:

• Power users who want Vim/Emacs-style keyboard-only operation
• Team-level collaborative documents (it positions itself as a personal writing tool)

Quickest way in

Grab the installer for your platform from Releases:

https://github.com/alexishida/Moji/releases/download/v1.0.7/Moji.Setup.1.0.7.exe

DMG, AppImage and deb builds are on the same page for macOS/Linux. After installing, double-click any .md file to open it, and switch between Edit / Preview / Split modes from the top toolbar.

Ctrl + \

That shortcut toggles the split preview, and the left/right pane ratio is remembered automatically.

Source: https://github.com/alexishida/Moji

SteveTheKiller/KillerPDF — a free editor with its own PDF 2.0 engine

Software that can edit PDFs on Windows is scarce to begin with, and free options are scarcer still. KillerPDF is an open-source take on it: it bundles PDFium rendering plus a PDF 2.0 / PDF/A / PDF/UA engine (shipped separately as the KillerPDF.Engine .NET library you can reference on its own), and supports text annotations, drawing, signatures, form filling, merge/split/rotate/crop/reorder, side-by-side comparison of two PDFs, and built-in Tesseract OCR (offline, with extra language packs available for download). All processing happens locally — no telemetry, no accounts.

What's more hardcore: every release runs regression tests against veraPDF's compliance corpus of 2,900 PDFs, and zero regressions is a hard requirement. That's a solid engineering attitude. It ships both a standard installer and a portable build, and the portable one bundles its runtime, so you can drop it on a USB stick and run. The CLI is complete too — every core operation runs headless from a terminal.

Who it's for:

• Windows users who handle PDFs (contracts, scans, forms) day to day
• People who don't want an Adobe Acrobat subscription and are uneasy about online PDF tools' privacy
• Anyone who wants to batch-process PDFs from the command line

Who it's not for:

• macOS / Linux users (only Windows builds for now)
• Enterprise teams that need cloud collaboration and multi-user annotations

Quickest way in

Download either the installer or the portable build from Releases, open it and drag files into the window. Batch merging from the command line is just as easy:

KillerPDF.exe --merge out.pdf a.pdf b.pdf scan.jpg

To turn a scan into a searchable PDF, use the OCR subcommand:

KillerPDF.exe --ocr scan.pdf searchable.pdf --lang eng

==Note: extra OCR language packs are downloaded on demand, so the first use needs an internet connection.==

Source: https://github.com/SteveTheKiller/KillerPDF

zarazhangrui/frontend-slides — let your coding agent build the deck

This repo is already close to 30k stars, and its form is unusual: it's a skill for coding agents, not conventional software. Install it into Claude Code, Codex, Kimi Code, Cursor, Aider or any agent that can read files and run commands, and all you have to say is "build me a pitch deck for an AI startup". The agent first generates 3 visual style previews for you to pick from, then produces the full HTML deck.

Each deck is a single HTML file with inline CSS/JS — zero dependencies, zero npm, zero build tooling. The selling point is "anti-AI-slop": it ships a curated set of design styles specifically to avoid the generic "purple gradient on white" AI aesthetic. It can also convert an existing .pptx into a web version, keeping all images and content.

Who it's for:

• People already using a coding agent for code who want it to handle presentations too
• Non-designers who need external materials that don't smell like AI
• Anyone who wants to turn a PPT into a hostable web page that's easy to share

Who it's not for:

• Pure Office users who don't use any coding agent
• Presentation designers who need pixel-level polish inside PowerPoint

Quickest way in

Claude Code users install it via the marketplace source. These two commands must be sent separately:

/plugin marketplace add https://github.com/zarazhangrui/frontend-slides

/plugin install frontend-slides@frontend-slides

For other agents (Codex, Cursor and so on), just send the repo link:

Please use this skill: https://github.com/zarazhangrui/frontend-slides

Once installed, state your requirement:

I want an 8-page technical talk on edge computing, with a calm visual style

The agent generates 3 visual previews for you to choose from, then produces the complete HTML.

Source: https://github.com/zarazhangrui/frontend-slides

4. Wrap-up and what to install first

If you can only install two, I'd go with RenewHelper + frontend-slides. The first solves the "domain/service silently expired" pain the same day, and the second turns the coding agent you already have into a slide designer — the best return on effort here. Go deeper on the other four as needed:

• Several VPS instances to monitor in one place → install CF-Server-Monitor
• Writing Markdown docs → try Moji
• Heavy PDF user → install KillerPDF
• Fed up with PowerPoint compatibility → give Bento a shot

#AIVideoEditing #ScriptedVideo #TalkingHeadVideo #Renewhelper #CfServerMonitor #Bento #Moji #Killerpdf #FrontendSlides #cf-server-monitor