2024 - Metaplex introduces Core, a new digital asset standard with unprecedented flexibility
2026 - Metaplex announces an Agent Identity layer with MIP-14
Could be something...
github.com/metaplex-fo...
2021 - Metaplex introduces Solana NFTs, inspired by ERC-721
H1 2023 - Metaplex introduces pNFTs, making Solana the first chain to bring back royalty enforcement
H2 2023 - Metaplex introduces Compressed NFTs, making Solana the cheapest chain for minting NFTs
10/
Year 5 is about execution and expansion.
More standards.
More composability.
More weird experiments that look unnecessary… until they’re obviously inevitable.
Still building.
9/
We’re still early.
Digital assets are becoming infrastructure for agents, games, communities, finance.
Metaplex is quietly sitting at the center of a lot of that.
8/
Personally, these 4 years reshaped how I build:
Design for extension.
Assume adversaries.
Optimize for clarity so humans and agents can reason about your code.
Clean primitives > clever hacks.
7/
The fun part? Watching builders extend it in ways we didn’t predict.
That’s when you know you built something real.
When the ecosystem surprises you.
6/
The Core plugin model changed how I think about digital assets.
An asset isn’t just metadata.
It’s a programmable surface.
Identity. Execution. Delegation.
Assets that can act, not just exist.
5/
What I’m proud of isn’t a single feature.
It’s that we’ve stayed stubborn about a few things:
– Credibly neutral standards
– Open composability
– Minimal guardrails early so builders can explore
– Real performance on real chain conditions
4/
A lot of my time has gone into the unsexy parts:
Edge cases.
Authority models.
Lifecycle validation.
Making sure extensions compose without blowing up.
No one tweets about validation logic.
But that’s where the leverage is.
3/
The shift that mattered most: moving from hype cycles to infrastructure.
Hype burns hot.
Standards compound.
If you get the primitives right, everything built on top moves faster.
2/
Back then, “NFT infra” mostly meant minting JPEGs fast.
What we’ve built since is very different:
• Core asset standards
• Compressed NFTs at real scale
• Launch primitives for fair distribution
• Extensible plugins that let assets evolve over time
1/
4 years at Metaplex today.
When I joined, Solana NFTs were still a live experiment. Standards were fluid. Tooling was duct tape and conviction.
It felt unstable.
That was the appeal.
CI is part of your attack surface.
If it has deploy keys, signing keys, API tokens, or publish credentials, treat workflow changes like production code.
Don’t rely on vibes. Design your trust model deliberately.
If your repo auto-runs CI with secrets after someone’s first merge, you’ve created a trust escalation step.
That step is attackable.
The old heuristic was: low-effort first PR = suspicious. At @Metaplex we pretty much immediately close typo fix PRs.
But AI just broke that signal.
Now anyone can generate a clean, high-quality first contribution in minutes.
For those who don't, the flow is simple:
• Submit tiny PR (typo/docs)
• Get merged
• Gain trusted status
• Submit workflow change
• CI runs with secrets
• Secrets get exfiltrated
This has happened many times to good teams.
PSA for open source teams using GitHub CI:
“Harmless” first PRs can be part of a real attack pattern. Most of you already know this and require approval to run actions for first-time contributors.
Confession: I tried Codex and was not impressed.
Where's your assembly god now?
The dev died, and you're going to give up on the token?
Amateur
Michael Crichton published a book in 2024...
...16 years after he died.
Learn to commit
That’s the real unlock.
Crypto doesn’t make AI smarter.
It makes AI bounded, inspectable, and accountable.
Intelligence without constraint is chaos.
Constraint without intelligence is dead code.
Put them together and you finally get something usable.
Take @Metaplex.
An AI can propose a mint, update metadata, manage ownership, coordinate creators.
But the protocol enforces the constraints. No hallucination survives contact with onchain rules.
AI plans. Protocols enforce. Clean separation.
That’s what audited, onchain protocols provide.
They don’t care how confident an agent sounds.
They don’t infer intent.
They don’t get socially engineered.
They execute if rules are met — or nothing happens.
AI + crypto works for one simple reason people keep missing:
AI needs rails 🛤️
AI guesses, hallucinates.
That’s fine for writing. It’s unacceptable for execution.
If an AI agent is going to touch assets, ownership, or permissions, something in the stack has to be allowed to say no.
Feedback or suggestions are welcome, as always!
If you want to take a deeper look, or extend or contribute to the project, the repo is here: github.com/blockiosaur...
This project was heavily inspired by @gamemakerstk's Platformer Toolkit.
While my version isn't nearly as pretty, I wanted something that could be bundled with Phaser classes and plugins and that could be directly imported to future projects.
Under the hood:
• Phaser 3 for the game layer
• React / Next.js for the control surface
• Finite state machine per movement state
• Central parameter registry + event bus
• Export or import directly to JSON
Everything is wired for instant iteration. No reloads.
Every parameter includes context:
What it does, why it exists, and what breaks if you push it too far.
The goal isn't just tweakability, but building intuition you can reuse elsewhere.
Just click the info icon next to any setting or category to see details.
I realized you can't just vibe code good game feel, but you can vibe code a tool to get you there.
Introducing the Platformer Game Feel Tester
ptest.breadheads.dev/
25+ parameters you can tweak in real time, 8 different presets based on popular games.
Change a value → feel it immediately