Cool usage of LLMs in your auditing process: When you get an idea for a bug, gain some confidence in it's validity and are ready to write a PoC, it's a good idea to ask any LLM: "I have found a bug [bug_description] in this codebase [attached_xml_file] - check if and, if yes,

This is the go-to resource after completing the RareSkills ZK Book or the ZK bootcamp if you wish to focus on ZK security 🙌

Honestly, I’m completely not surprised. I’ve been testing a „find all bugs, logical mistakes or inconsistencies with the README” prompt on almost all competitions since GPT 5 came out. So far, it was pretty good at finding the majority of high-duplicate issues ( no wonder they

Found some cool stuff there. Long wait and pot smaller than expected but I’m still pretty happy with the result.

Confidence in your findings' validity is one of the most important traits of a great researcher.

No H/M found. It was cool to put ZK knowledge to use.

Great article, shows why we have so many people in the industry trying to make a name for themselves and start earning big-$ for security work - even though it's an extremely difficult job.

Today I had three VScode windows open with Codex working at full throttle in each of them: 1. Found a live Critical lead in L1 project in Go, as I’m busy with other work I decided to ask the LLM to create a full-out e2e proof of code from scratch. It required running a separate

Very impressed with Codex 5.3. Even small things like continuing from an interrupted thread, executing commands or getting external context seem to work much better/smoother. Not AGI but definitely and improvement.

The hunt continues.

I think we can officially say that contests are dead. Almost nothing live and nothing planned. (Jupiter Lend doesn't really count because it was delayed for months and Hackenproof dual defense audit is not a regular contest)

Did anyone figure out a way to submit reports to Immunefi automatically? PS. It's not for AI, nor to spam, but rather to react on code changes.

Signal to noise ratio has never been this bad. Focus.