Small tools that do something real.
Ten working instruments across security, machine learning and AI, not screenshots or slideware. Paste something in and watch it happen: an attack fooling a neural network, a phishing link unmasked, a log stripped for indicators, a model rationalising a biased answer, ignoring its own reasoning, or narrating a cause you can check. Every one runs entirely in your browser.
- Runs in your browser
- Nothing you type is uploaded
- Each shows its own working
Adversarial examples playground
Fool a neural network in real time. Draw or pick a digit, add an imperceptible perturbation, and watch the classifier flip while the image looks unchanged, a live, from-scratch FGSM/PGD attack. This is the idea at the centre of my research; if you try one thing here, try this.
Open the playground →- CH-8
Unfaithful reasoning
Add a bias a model is never told about and its answer moves, but its chain of thought never admits why. An interactive reproduction of a known result: stated reasoning is not the same as actual cause.
Open → - CH-9
Load-bearing reasoning
Break a model’s chain of thought, truncate it, corrupt a step, or swap it for filler, and see whether the answer moves. Often it does not: the reasoning was decoration. A reproduction of the chain-of-thought perturbation tests.
Open → - CH-10
Faithfulness, answer key held
A deterministic scorer ranks OSINT findings and an LLM only narrates, so the true cause is known exactly. Toggle a signal and watch whether the explanation tracks it, confabulates, or omits. Built on Glean’s design.
Open → - CH-1
Live global attack traffic
See which ports the internet is attacking most right now, live from a worldwide honeypot network, and what each attack actually is.
Open → - CH-7
Phishing URL inspector
Paste a link and see where it truly points. Flags lookalike domains, brand impersonation, typosquats and the classic tricks, without ever opening it.
Open → - CH-4
IOC extractor
Drop in a log or a phishing email and pull out the IPs, domains, URLs, hashes and CVEs, de-duplicated and defanged, ready for a ticket.
Open → - CH-2
Password strength lab
Not a red-yellow-green meter: the actual entropy, the patterns an attacker exploits, realistic crack times, and an optional breach check.
Open → - CH-5
CAN frame decoder
Decode a raw automotive CAN-bus frame into its ID, type, length and per-byte view, the protocol my adversarial-IDS research defends.
Open → - CH-6
Hash & encoding workbench
SHA-1/256/512, Base64/hex/URL encode-decode, and a JWT decoder, the everyday conversions, without pasting secrets into a random site.
Open →
Built by Yulan Galagoda. No account, no cookies, no tracking inside the tools, the code for each is on the page it explains.