Skip to content
A stock WordPress exposes 133 REST routes, 80 XML-RPC methods and 1,514 PHP files; this static site runs nothing on request

A Static Site Has Nothing To Break Into

This blog is built with Hugo and served as 6,343 plain files, with no code of mine running on any request. Against it, a stock WordPress 7.1.3 measured in a podman lab: 133 REST routes, 62 of them accepting writes, 80 XML-RPC methods, 1,514 PHP files, every one of its 3,785 files writable by the web server’s own user, and a database account holding every privilege. This post counts both attack surfaces, explains what Hugo’s own security model does at build time, owns up to where the risk moves once nothing runs at serve time, including two gaps in how this site was built, one of them now fixed, and sets out how to build a static site offline, once, and upload nothing but the finished files.

23rd October 2026 · 17 min · 4046 words · Damien Dye
757 PHP interpreter CVEs since 1997, at least 288 memory corruption, 172 rated full remote compromise without a login

PHP: Twenty-Nine Years Of Leaving The Door Open

PHP2Shell, disclosed on 21st October 2026, is unauthenticated remote code execution in the PHP interpreter itself, every version back to 5.4, exploited before the patch existed. It is the latest entry in a long record: 757 CVEs against the interpreter since 1997, at least 288 of them memory corruption and 172 rated as full remote compromise without a login. This post counts that record from NVD and the CISA KEV catalogue, explains why a C engine and a language that took a decade to remove register_globals make the same mistake twice, looks at what WordPress does to the numbers, how web shells stay put and how to stop them, the two times php.net itself was broken into, the share of the web still running versions nobody patches, and why the usual deployment, one process holding the templates, the logic, the SQL and the database password, turns every one of those bugs into everything. And why, having given up on PHP myself, I think you should look at leaving it for a memory-safe language such as Go.

23rd October 2026 · 29 min · 8074 words · Damien Dye
A phone holding a certificate from your own CA passes nginx or Caddy at home and reaches Home Assistant, while a scanner without one is refused in the handshake

The Home Assistant App Locked To A Certificate, With nginx Or Caddy And No Cloudflare

Locking the Home Assistant app to a client certificate with nginx or Caddy at home, with step-ca as the certificate authority instead of Cloudflare. What changes when Cloudflare leaves the path, an nginx config and a Caddy config each tested with no certificate, a good one, a revoked one and a WebSocket upgrade over IPv4 and IPv6, why nginx can check a CRL and Caddy can only keep an allowlist, the Home Assistant trusted-proxy settings, what the phone needs installed, and where it falls down.

9th October 2026 · 10 min · 2147 words · Damien Dye
A phone holding a client certificate passes Cloudflare's edge check and reaches Home Assistant through a tunnel, while a scanner without one is blocked at the edge

The Home Assistant App Over Cloudflare, Locked To A Certificate, On The Free Plan

Reaching Home Assistant from the Android app away from home without putting a login page on the internet. How the remote-access options compare on exposure, real cost (machine load and DDoS as well as the bill), blocking risk and battery, why Cloudflare Access and service tokens do not suit the app, how the free-plan client certificates and a WAF custom rule lock a tunnel hostname to the phones that hold a certificate, which Home Assistant settings the reverse proxy needs, building the PKCS12 file, putting a Let’s Encrypt certificate on the origin and why renewal needs a restart, what the scanners did in the first two hours, and what the phone, Chrome, Helium and the Companion app’s two clients each did with the certificate. A self-hosted nginx or Caddy version is a separate post.

9th October 2026 · 19 min · 4709 words · Damien Dye

My YubiKey Logs Me In. My Keyring Never Got Told.

Log in with a YubiKey and GNOME asks for a password anyway, because the login keyring is encrypted with a password your login never produced. This post covers why pam_u2f being sufficient starves the keyring, what the usual fixes cost (blank password, typing it anyway, the TPM), and the fix I built: the key’s FIDO2 hmac-secret, gated on PIN and touch, decrypting a stored copy of the account password and unlocking oo7-daemon at login. Then what the build turned up: Fedora 45 had swapped gnome-keyring for oo7, oo7’s placeholder keyring takes the first password you type as its own, the FIDO2 tools only read a PIN from a terminal, and a reset script that would have left me with no keyring at all, caught by testing it against a throwaway daemon first. The scripts, an installer and that test harness are available to download.

7th October 2026 · 12 min · 2567 words · Damien Dye
A chain of trust running from the signed root through a signed TLD and stopping at an unsigned second-level domain, with the counted figures beside each step

DNSSEC: Protecting Your Traffic From Forgery

The hard part of DNSSEC was finished years ago. Counted from the live root zone on 27th September 2026, 1,351 of 1,438 top level domains carry a DS record and every one of the 1,038 gTLDs is signed. Then it stops dead. A census of every gov.uk domain in the official register finds 39 signed out of 2,390 that still resolve, nine of them parish councils, while HMRC, the NHS, GCHQ and the National Cyber Security Centre are not among them. One certificate authority in nine has signed. So has one Linux or BSD distribution in five. windowsupdate.com has no DS at all. This is what a forged answer actually costs, what signing does about it, why the usual excuses do not survive contact with the numbers, and whether forty-two years after Mockapetris the real problem is that almost nobody understands what DNS actually promises.

27th September 2026 · 48 min · 10979 words · Damien Dye

DNS Over HTTPS Walks Straight Past Your Controls

DNS over HTTPS encrypts your lookups, which is good, and sends them to a resolver of the client’s choosing on port 443, which is the problem. Your own resolver never sees the query, so the blocklist it would have failed, the threat feed it would have hit and the log line it would have written all vanish. This post walks the mechanism: why one encrypted web connection among thousands is invisible at the border, who on a machine can pick a resolver you never chose (the OS, the browser, any app, any script on any web page, and malware), and the documented cases, from a page’s own JavaScript querying a public DoH endpoint that allows cross-origin requests, to Godlua and PsiXBot hiding their command channel inside DoH, to the OilRig APT exfiltrating data over it. Then the case that the privacy story is a cover: DNS over TLS already encrypted DNS in 2016, on a port the network operator can still govern, so the only thing DoH adds over it is defeating the admin, which is why the browser vendors who wrote and shipped it, and the advertising firm that runs the biggest public DoH resolver, are the ones who gained. Then the fix, which is not to ban encryption. Run your own DoH and DoT resolver, advertise it with Discovery of Designated Resolvers, refuse port 53, port 853 and known public DoH resolvers at the border, and answer the Firefox canary so the browser stands down. Keep the encryption. Take back who chooses the resolver. It closes on the fallout: the same DNS-based enforcement runs across Europe and in Australia, and courts in Italy, France, Belgium and Germany are now ordering the public resolvers themselves to block, with Cloudflare fined and appealing, Google refusing in silence and OpenDNS switching itself off for whole countries. The pattern underneath is a US design imposed on the world, and a US government that casts other countries’ laws as extortion of American firms and warns them not to comply.

24th September 2026 · 38 min · 9604 words · Damien Dye

Your Firewall Takes Instructions From Strangers. Turn Off the Protocol Helpers.

A protocol helper — SIP ALG, FTP helper, H.323 ALG, conntrack helper, call it what your vendor calls it — reads the payload of a connection, finds an address and a port written in the text, and opens an inbound pinhole for them. It cannot tell whether that text came from a real FTP client or from a hidden form on a web page, because there is nothing in it to tell. Samy Kamkar proved the browser case in 2010. NAT Slipstreaming proved it again in 2020, and Armis extended it in 2021 to reach any device on your network, not just the machine that clicked. The IETF asked for these off by default in 2007, Linux turned them off in 2016, and the browser vendors ended up shipping a blocked-port list that reads like a directory of conntrack modules. This post walks the mechanism diagram by diagram — the expectation table, the segment-alignment trick, H.323 call forwarding, the IRC helper that fires on somebody else’s message — takes in the IPsec pass-through helper, which cannot read ESP at all and steers inbound packets on an SPI it watched go past in the clear, sets the lot against the Cyber Essentials firewall control it plainly fails, and gives the commands to turn it all off on Linux, Cisco, Juniper, FortiGate and MikroTik.

14th September 2026 · 54 min · 12564 words · Damien Dye

What Your MSP Built You, And Who Else Can Reach It

Part 2 of 3. What actually gets built once the paperwork is signed: the box they will not be talked out of, the firewall that is the way in, the basics that were the thing you bought, and the agent on every machine that answers to somebody else’s console.

28th August 2026 · 39 min · 8266 words · Damien Dye

Backdoors, and Who Gets Accused of Them

Shutting a supplier out on security grounds needs a standard, applied evenly. Part 6 of 8 sets out what has been established about weakened products and interception, including a Swiss parliamentary inquiry, and shows why a built-in way in belongs to whoever reaches it.

25th August 2026 · 6 min · 1099 words · Damien Dye