Launching today

xnftables
A deny-by-default firewall thought experiment and template
4 followers
A deny-by-default firewall thought experiment and template
4 followers
A deny-by-default Linux nftables thought experiment and template. A starting point for tightly scoped private assets, not a production-ready policy or security guarantee.






Free
Launch Team

AWS Activate CreditsCredits that fuel your startup’s next milestone
Promoted

1/5 · What if access started with a deliberate no?
xnftables explores that as a deny-by-default Linux nftables template. It is mostly a thought experiment, not a policy I'd tell anyone to drop into production. At scale it's technically difficult, so a better starting point is a tightly scoped personal or confidential asset, tested in a lab with recovery access and a rollback plan.
My cybersecurity background made me want something sysops and engineers could read, question and adapt, rather than inherited rules whose purpose had long disappeared. The discussion it starts may matter as much as the configuration. Starting from a closed policy means every open port has to be justified in the rules. That habit of asking why something is allowed is the part I hope people take away.
Today is Day 9 of 11 of the Product Hunt Marathon: 11 free, open-source launches, one a day, 10 to 20 November. xnftables is dedicated to my personal hacker wall of fame: Aaron Swartz, Jonathan James, Bassel Khartabil and Barnaby Jack.
Where does the policy become impractical, and what would you change before trusting it with an asset you care about?
2/5 · My personal hacker wall of fame
xnftables, Day 9 of 11 of the Product Hunt Marathon, is dedicated to my personal hacker wall of fame: Aaron Swartz; Jonathan James, "c0mrade"; Bassel Khartabil, also known as Bassel Safadi; and Barnaby Jack.
Four different lives, not a shared endorsement or a single moral. Aaron's open-access work and Bassel's free-culture work make the sharing matter to me. Barnaby's research reminds me that a demonstration should help somebody repair a system. Jonathan's story makes me think about how we give curious people mentorship, accountability and constructive places to learn.
My small contribution here is a readable firewall template, not a comparison between my work and theirs. Test it in a lab, understand the rules, keep a rollback path, and share a sanitised improvement.
Read their histories: Aaron (https://www.eff.org/deeplinks/2014/01/remembering-aaron); Jonathan (https://www.wired.com/2009/07/hacker-3/); Bassel (https://creativecommons.org/tag/freebassel/); Barnaby (https://www.bbc.com/news/technology-23467411).
3/5 · Scope, limitations and safe testing
Megamesh (https://megamesh.sh) was the personal-VPN project on Day 7. Today, Wednesday 18 November, returns to cybersecurity with a separate project, not a Megamesh upgrade. It is Day 9 of 11 in the Product Hunt Marathon: AI Credit Checker was yesterday, and Timescreen follows tomorrow.
It's a Linux nftables policy template built around deny by default: public services require explicit exceptions. I used to work in cybersecurity and still do occasionally. I wanted a policy I could read and explain, rather than inherited rules whose purpose had disappeared.
This is technically complicated to deploy at scale. Real estates have service dependencies, exceptions, remote access, failure modes and people who need to operate them. I'm not presenting a template as an easy fleet-wide rollout.
I'd like it to give sysops and engineers a second thought about the access they take for granted. Even if you never deploy it unchanged, the policy can be a useful starting point for a tightly scoped personal server or an online asset intended to remain strictly confidential. A firewall alone does not establish confidentiality: identity, application security, secrets, updates and operations still matter.
For this launch, caution comes first. Adapt the policy to your topology, test away from production, verify recovery access and keep a rollback plan. A template is not a security guarantee; deployment deserves qualified review.
4/5 · Ownership and the work after launch
This is where the CHO narrative meets an operating responsibility. One lead or a small enabling team can own the tool programme, budget and handoff without pretending to replace security specialists. Let AI help inspect a policy; keep predictable packet enforcement in rules. Give qualified people lab tests, recovery access and a rollback plan.
The organisation should inherit a maintainable system, not dependence on its original builder or an agent improvising in production. Replacing a subscription only makes sense when the quality and full operating cost hold up.
That's how I intend to handle feedback across these projects. I'll consider each report and request, prioritise bugs and meaningful small improvements, and do my best to fund worthwhile AI-assisted iterations. I won't turn every suggestion into a feature.
Keep production policy, tests and change history in organisation-controlled Git, preferably infrastructure you operate properly and back up. Keep private topology, keys and incident details private. Share reusable fixes and sanitised tests upstream so other operators and their AI assistants can build on reviewed work. Validate every fork in its own environment.
5/5 · Contribute, or get help with your stack
A useful report includes a sanitised rule, the traffic you expected to pass or fail, and a reproducible mismatch. If you use Astra or your AI, ask it to explain the current rules, propose one bounded change within your budget, then test both allow and deny cases. Check the actual behaviour, not just the explanation.
The repository is MIT-licensed. You can extend or commercialise it with the notices retained. Operational responsibility remains yours.
If this points to a broader cybersecurity problem, or you need help with a secure stack, visit https://megastack.sh. I've brought together a group of friends whose specialist expertise I value highly.
Read the template: https://github.com/paulfxyz/xnftables
The wider experiment: https://ph.paulfleury.com/
Web Summit in Lisbon is over (9 to 12 November), and thanks to everyone I met there. I live in Lisbon and am happy to meet around the city. Tomorrow, Day 10 of 11 of the Product Hunt Marathon, is Timescreen, and Day 11 is Mars on 20 November. On Monday 23 November I'll publish a full report on LinkedIn and X on all 11 launches.
Separately from the marathon, I'm also working on a few ventures and looking for the right people. Openline (https://openline.com) is my main venture and startup, and I'm looking for VCs, angels and investors: the aim is to reshape the eSIM market, with a deep pipeline of ideas ready to execute. Iris (https://iris.online) and Digital Spot (https://digital.spot) are spin-offs of tools built for Openline. Iris is one inbox for customer messages across 19 channels, with AI-drafted replies you approve; Digital Spot is social media on autopilot, planned four weeks ahead, where an AI proofreads and only human-verified posts publish. I'm looking for investors and early customers for both.
Enki (https://enki.ngo) is rather a thought experiment: a non-profit, local-first AI public good, too big for me alone; I'm looking for philanthropists, donors and helpers. Cup of News (https://cupof.news) is a Sunday-afternoon hobby that is soon turning into an app, an open-source, self-hosted AI morning digest; I'm looking for contributors and users.
CHO (Chief Hacking Officer) is a deliberately playful title for serious consulting work. I'm open to a CHO brief of up to 30h/week with a great company or startup, a worthwhile product, trust and a modest budget. For any of this, reach me by DM on X (@paulfxyz) or LinkedIn, or at paulfleury.com/contact :)