Evidence Adejuwon

Software engineer · founder @codetribeio · shipping production AI products · open to senior full-stack / founding eng roles at AI-first companies

[Evidence Adejuwon @evidencecodes]

@lightsilver323 https://t.co/KaMuumJqZx

[Evidence Adejuwon @evidencecodes]

Really good point, and this is actually the direction I'm already leaning. The CLI is meant to support bringing your own model and provider, so you're not locked into one, and it also means nothing has to go through us at all if you point it at your own model. Likely open-sourcing the CLI itself for that reason. The sandbox execution and orchestration stays closed source though, that's the harder part to get right. Alongside all that, there's also a separate Semgrep and OWASP layer that doesn't depend on any model at all.

[Evidence Adejuwon @evidencecodes]

How much of your code was written by AI lately? Mine too.

Here's the problem nobody talks about: bugs that look fine, pass review, and ship anyway.

So I built Syntinel. An AI that reads your codebase like an attacker would, SQL injection, hardcoded keys, auth bugs, prompt injection, then writes the actual exploit and the fix.

One command. Real security audit. Not public yet, but here's a first look 👇

Comment "SYNTINEL" if you want in when this goes live.

[Evidence Adejuwon @evidencecodes]

@markproduct AI-powered security. Fix vibe-coded apps

https://t.co/KaMuumJqZx

[Evidence Adejuwon @evidencecodes]

@whotterre Join the waitlist 👇
https://t.co/KaMuumJqZx

[Evidence Adejuwon @evidencecodes]

@anupamrjp https://t.co/KaMuumJqZx

[Evidence Adejuwon @evidencecodes]

Most existing tools, Copilot Code Review included, read the diff and flag what looks risky, basically a smarter linter. Syntinel is meant to actually generate tests that try to break the function and prove it fails, not just guess. On large or nested repos, it's only meant to look at what changed, not the whole codebase, so depth shouldn't matter in principle, but I haven't stress tested a real large Kotlin/Java monorepo yet, that's high on my list.

[Evidence Adejuwon @evidencecodes]

Yeah exactly, that's the plan, though none of this is live yet, still early and in active development. For CI, generated tests would run in a secure, isolated environment before anything unsafe touches the pipeline. For the PR, before merge, a GitHub App would flag issues by severity and block the merge button on anything critical, not a suggestion. For the CLI, same core engine, running locally, so you catch things before it's even a PR. Right now I'm heads down on the GitHub App first.

On training on your code, fair question, and important for a corporate project. Nothing leaves your machine except what's sent off for the actual review, and the longer term plan is a fully self-hostable option so sensitive codebases never touch a third party at all. I'll have a clear, published data policy before this leaves beta, not just asking anyone to just take my word for it.

[Evidence Adejuwon @evidencecodes]

@AkohTech Join the waitlist

https://t.co/KaMuumJqZx

[Evidence Adejuwon @evidencecodes]

@OlaleyeOl12562 Join the waitlist

https://t.co/KaMuumJqZx

[Evidence Adejuwon @evidencecodes]

@sflorimm Claude $100

[Evidence Adejuwon @evidencecodes]

@JakeFleshner AI-powered security

[Evidence Adejuwon @evidencecodes]

@jayvraavi Great one man 🎉

Living most solo founders dream 💯

[Evidence Adejuwon @evidencecodes]

@AdityaShips Get a good VPS

You can get from Digital Ocean and Linode

[Evidence Adejuwon @evidencecodes]

@fidjissimo I wish you a speedy recovery. The world needs more of your touch.

[Evidence Adejuwon @evidencecodes]

Teaching a model to shop by sight 👀

Building visual product search for Zetix. Snap a photo, find the exact item. Off-the-shelf CLIP only hit 55.5% top-3 accuracy, so we're fine-tuning our own on real product images.

Watching the loss curve fall live... 3.25 → 1.16 and dropping. All training locally on a MacBook

[Claude @claudeai]

Michael Truell (@mntruell) fell in love with coding at 12. The company he co-founded, @cursor_ai, went from 15 people to 700 in two years.

Today, over 60% of the Fortune 500 build with its AI coding platform.

[Evidence Adejuwon @evidencecodes]

Interesting side effect of fable 5 being this good on claude code: opus 4.8 now feels like a downgrade.

hit 95% of my session after hours of work, swapped models to manage usage, and the quality dip was immediate and obvious. didn't expect the gap to be that visible in real work.

[Claude @claudeai]

Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.

Its capabilities exceed those of any model we’ve ever made generally available.

[Mr. Buzzoni @polydao]

https://t.co/E8GJ7yfZEM

[Evidence Adejuwon @evidencecodes]

@davidokocha086 This is solid work. Building it in C from scratch instead of reaching for a library is the kind of thing most people skip, and it’s exactly how you build real intuition. Keep shipping these in public.