Before shipping our public API, we audited our own product. Found: a live GitHub credential leaking on every user read, report tokens with cross-tenant write access, 11 vulnerable deps in the browser bundle. All fixed. Also: 9 new vulnerability classes, noise down 25%. 🧵👇
Full Blog:
fidesium.xyz/august-2026-tec…
The team keeps bringing the wins.
I just found a confirmed bug on @immunefi as a chief finder #immunefitribe immunefi.com/s/ss/?severity=…
Fidesium retweeted
🍌 24 HOURS TO GO until Ep.2 of the BananaZone Weekly 69-er with special guests:
@HelloBenWhite from @predictfully_ ,
@merheb from @EYEKONSTUDIO ,
@TonyCatoff (The Puja King of @solflare ) and
@bockus from @Fidesiumapp
The leader board goes live for 69 minutes on desktop and @solanamobile
This week the top 10 get rewarded 🍌🍌🍌
Every founder asks what an audit costs.
Almost nobody asks how long it lasts.
Kind of strange don’t ya think, because you'd never buy anything else this way. You'd never lease equipment without asking about depreciation. You'd never buy insurance without checking the coverage period.
But teams spend $80k on an audit and never ask the only question that determines its value:
What is this worth after my next release?
Because an audit doesn't have a price. It has a depreciation curve. It's worth the most on the day it's delivered and loses value with every merge, every dependency bump, every config change, every new integration.
This is what I call Audit Durability. How long an audit's assurances actually survive contact with your roadmap. What can we as a team do to extend this coverage? And for most fast-shipping teams, the honest answer is quite a lot with the correct framework.
Now run the real maths:
$80k audit ÷ 12 months of assumed coverage = feels reasonable.
$80k audit ÷ 6 weeks of actual Audit Durability = the most expensive PDF you own.
The audit industry sells you the first calculation.
Your attackers are working off the second one.
Fideium July 2026 Update
July was all about accuracy infrastructure.
We built the measurement system the product needs to improve reliably, then used it to cut noise by 20% and ship 16 new vulnerability classes.
Here’s what shipped:
We can now prove our accuracy claims.
Moved from a single 143-file academic benchmark to four corpora covering ~63,000 contracts and ~6,200 labeled findings from real paid audits (Code4rena, Sherlock, DAppSCAN). 423 individual projects now carry accuracy floors enforced on every pull request.
Signal-to-noise improved 20%.
Across 606 real audited protocols, unmatched findings on the DAppScan benchmark dropped from 22,893 to 18,269. Alarm fatigue is the #1 reason security tools get switched off — so we treated it as a first-class problem. Several detectors improved by more than 90%. We also retired five detectors that only added noise with no unique coverage.
16 new vulnerability classes, including strategically important ones:
Proxy storage-slot collision (governance-takeover risk)
Oracle single-point-of-failure
A decomposed five-detector denial-of-service family that replaces one imprecise catch-all
Two of the new detectors found real bugs that the original human audits missed.
Next month
Continued precision work on the highest-noise detector families, promotion of the proxy-collision suite from gated to live, and closing the recall gap the Sherlock corpus exposed.
Building the foundation that lets accuracy compound.
12 hours → 5 minutes.
Balancer scan, before and after Q1.
Also shipped:
• Cross-contract taint analysis
• On-chain bytecode scans, 30+ EVM chains
• Solana fuzzer, 5x faster
• 90%+ better reentrancy detection
Changelog: fidesium.xyz/q1-2026-platfor…
Security and compliance are what we do.
Honoured to be a part of shaping London's crypto future.
The team audited 36,000 lines of Rust on @AdrenaProtocol .
The findings weren't where most auditors look.
Here's the framework that surfaced them.
Great day with @SuperteamUK
Audited and still hacked: $600M+ lost in 2026. 5 op-sec rules, 6 technical questions to ask w/ @Fidesiumapp nitter.cf/i/broadcasts/1jxXgeZov…
Security day at @SuperteamUK BuildStation
How to think adversarial on Solana.
@cryptoabe4
Our co-founder and CTO @cryptoabe4 will be at @consensus2026 Miami, May 5–7.
If you want to talk #smartcontractsecurity, #automatedaudits, or #devtooling, reach out to him directly at @cryptoabe4.
See you there.
🆕 What We Built in Q1 2026 - And Why It Matters
Most #smartcontractsecurity updates read like patch notes. Here's ours framed around the problems these improvements actually solve.
⚡️ 140x Faster. Zero Quality Trade-Off.
Our scanning engine got a ground-up rebuild this quarter. Large repository scans - like Balancer - dropped from 12+ hours to under 5 minutes through deep algorithmic and memory optimisations.
The important part: detection quality didn't move. Every optimisation was validated against our full benchmark suite before shipping. Faster scans that miss vulnerabilities aren't faster scans; they're false confidence.
—
🪲 Cross-Contract Bugs Are No Longer Invisible
Most scanners work file-by-file. Real exploits don't.
Fidesium now supports multi-file scanning with full cross-contract flattening and deduplication. The scanner resolves import trees, traces taint across contract boundaries, and surfaces bugs that live in the interaction between contracts, the ones that actually get exploited in production.
—
⛓️ On-Chain Scanning Across 30+ EVM Chains
There's a gap between the code in your repository and the bytecode deployed on-chain. Fidesium now lets you register deployed contract addresses and scan the actual bytecode directly. No repo access needed. No assumptions about whether what shipped matches what was reviewed.
—
🆙 A #SolanaFuzzer That's 5x Faster Than Public Alternatives
We built a high-performance #Solana fuzzing harness, natively bridging CPI via LiteSVM onto native hardware. In benchmarks, it outperforms publicly available fuzzers by approximately 5x in both operations per second and memory efficiency.
Deeper fuzzing. Faster engagement spin-up. State spaces that were previously cost-prohibitive to test.
—
📃 90%+ Better Reentrancy Detection
We overhauled our reentrancy detector and validated it against the SmartBugs curated benchmark. It now handles cross-function reentrancy paths, self-call guard patterns, and pre-Solidity 0.6 syntax - the variants most automated tools miss entirely.
Across the board, we ran a systematic false positive reduction effort. Better cross-contract taint analysis, improved call graph resolution, and recognition for patterns that previously generated noise. The goal: findings your team acts on, not ones they dismiss.
—
The Bigger Picture
Every improvement this quarter reflects the same thesis: the future of #web3security isn't choosing between human expertise and automation. It's building infrastructure where they compound each other.
That's what we're building at Fidesium.
Full technical breakdown on the blog → buff.ly/nYSeeXm
#Audits are signals.
Not systems.
A signal tells you: "This was secure at this point."
A system tells you: "This is still secure now."
Most teams confuse the two.
That's not a process problem.
It's a mental model problem.
#web3security
Last week at the @SimplicityWeb3 Group accelerator, one thing stood out:
Founders are asking better questions.
Not just:
“How do we grow faster?”
But:
“What assumptions are we building on?”
Our co-founder @bockus joined a strong group of speakers alongside teams from Base, DWF Labs, Canton Network and others, sharing a perspective that cuts across both #startups and #Web3security:
Shout out to what Simplicity Group is building here.
More rooms where founders think at this level.
#FirstPrinciples
Excited to announce that we
will be attending EthCC[9] in Cannes!
March 30 – April 2 at the Palais des Festivals.
DM us, let's talk code, audits, and keeping things safe at scale. See you on the French Riviera!
Security team trying to affect real change in the ecosystem. You should check them out. The team also let me know they will be showing the film at @EthCC
We are really proud to support @ChainPatrol and their project on @Giveth! 💜 Read about their impact or support them here:
giveth.io/project/lights-in-… #Giveth






















