Anon Jon

Software Engineer building @orlojHQ
Co-Creator @ATXDAO

[Anon Jon @An0n_Jon]

If you're building production agent pipelines and hitting this, curious what you're doing about it.

[Anon Jon @An0n_Jon]

Every major agent framework makes the same architectural mistake with MCP servers. 🧵

[Anon Jon @An0n_Jon]

This is one of the new features we've built in Orloj. MCP servers as ephemeral containers. https://t.co/7rpJZ6fnvh

[Anon Jon @An0n_Jon]

@ThePrimeagen https://t.co/NNaDLUs3Fh
Agent-infrastructure-as-code

[Anon Jon @An0n_Jon]

@JamieMallers Yep emits tracing for every step in the task lifecycle so you have a full audit trail for agent, tools, etc

[Anon Jon @An0n_Jon]

Running AI agents in prod today looks like running containers before Kubernetes, ad-hoc scripts, no governance, no observability.

So I built Orloj: an open-source orchestration runtime for multi-agent AI systems.

Declare your agents, tools, and policies as YAML. Orloj handles the rest.

→ DAG-based orchestration (pipeline, hierarchical, swarm)
→ Model routing across OpenAI, Anthropic, Azure, Ollama
→ Tool isolation in containers/WASM
→ Governance built in, policies, roles, permissions enforced at runtime
→ Idempotent replay, retry, dead-letter handling

GitHub: https://t.co/0JsCIrPyEB
Docs: https://t.co/YZSo6IsRjU

[Anon Jon @An0n_Jon]

Orloj has been running in the wild for a bit now and I'd love to hear from you.

If you've tried it, what was the DX like? Did the YAML feel natural? Anything missing that you wished existed?

I'm also actively looking for contributors. Whether it's a new feature, a rough edge you hit, or docs that confused you, all of it is welcome.

This is early days and the people who show up now genuinely shape what Orloj becomes.

Drop a reply, open an issue, or just star the repo if you want to follow along.

https://t.co/0JsCIrPyEB

[Anon Jon @An0n_Jon]

@adreon @mxlcolm hide your kids hide your wife

[Anon Jon @An0n_Jon]

Been seeing a ton of geospatial app demos lately.

Made me want to build one myself but for something meaningful.

I’m a dad, so I built a map of every registered sex offender in the U.S.

Public data should be easy to access when it comes to your kids.

https://t.co/c5BTkkwI6q

[Anon Jon @An0n_Jon]

@0xGooseOps This is the interesting part, every state has some weird API integration that is all rate limited and IP limited. I was able to get AI to make a sneaky script to scrape everything and then my VPN IP change when I got blocked. Like they dont want me to get the data or something...

[Anon Jon @An0n_Jon]

I can tell you're a good dude if you fully close the sauna door when you leave.

[Anon Jon @An0n_Jon]

The concept of randomness at a low-level is a very interesting rabbit hole to dive into. What is truly random? How do you integrate randomness into your contracts without needing a ton of third party dependency contracts? Our concept for time-series randomness is a novel approach to achieving randomness just in solidity thats provable via ZK. Test it out today in our (definitely) not a casino app.

[Drosera @DroseraNetwork]

The Drosera team is always trying to think of innovative features that traps can enable. Recently, we spent some time in the Greenhouse brainstorming and stumbled on a new discovery:

Traps can perform verifiable random number generation.

On-chain randomness is difficult to perform and requires you to think a bit more philosophically about what randomness actually is and how you get a random number in a trust-minimized way.

At first we weren't sure if it was even possible to create a pseudo random number given all the challenges... but we had a lightbulb moment.

The interesting thing is that it's possible to do this using one of the Trap’s super powers.

Time-series analysis.

This allows for a trap to use data over time to create a psuedo-random number that cannot be guessed beforehand, only after the number has been generated it can be proven it was generated deterministically.

This means that a Trap can perform random number generation that can be proven with ZK.

Random numbers can be used for tons of applications ranging from fair leader selection in consensus mechanisms, fair minting processes, random challenges in zero-knowledge proof systems, random sampling for governance, fair lotteries, and even gaming.

Check out our new roulette POC to see it in action 👇

🔗 https://t.co/3NNYsBpQrB

[Drosera @DroseraNetwork]

The first Drosera workshops made one thing clear:

This community is ready to build.

The workshops sparked curiosity, but many asked for more concrete examples to see what's truly possible. We heard the feedback and put together 10 Trap examples for DeFi automation.

Full project structure, Forge tests included:

https://t.co/izs2MHOcqX

These examples were made to serve as seeds for your imagination. To spark new ideas and discussions. Drosera’s magic lies in how easy it is to build and how quickly the community can turn small ideas into real infrastructure.

So if you believe in Drosera, plant something in the Greenhouse.

Grow an idea.

Cook something no one’s seen before.

Traps aren’t just for security. They can automate DeFi, enhance NFTs, upgrade governance, reshape trading, and more. Just check out some of the community examples below:

Gas fee monitor
https://t.co/VEvaBN43eC

Traps with NFTs
https://t.co/yIY6T4QiVn

Game of traps
https://t.co/0fchkLNkI7

A society grows great when old men plant trees whose shade they know they shall never sit in.

Let’s plant those trees for Ethereum.

[Anon Jon @An0n_Jon]

Holesky sucks now, come to hoodi.

[Drosera @DroseraNetwork]

The Ethereum Foundation announced the winding down of the Holesky testnet.

What does this mean for projects like Drosera?

Holesky support will continue until September 2025, but the time to look ahead has already begun. You can read the full announcement from the Ethereum Foundation here:

https://t.co/2FkbPF0Ddg

For Droserans, Holesky served as a proving ground for our earliest Trappers and Operators. Together, you deployed 33,756 traps and nearly 25,000 nodes. Helping shape a foundation for decentralized automation.

The next step?

Migration to Holesky’s successor: Hoodi.

We’ll be continuing our journey on Hoodi, with even more activities, challenges, and new ways to build using Drosera.

If you want to stay involved, follow the guide below:

https://t.co/hqyAXcoTHr

[Drosera @DroseraNetwork]

The Ethereum Foundation announced the winding down of the Holesky testnet.

What does this mean for projects like Drosera?

Holesky support will continue until September 2025, but the time to look ahead has already begun. You can read the full announcement from the Ethereum Foundation here:

https://t.co/2FkbPF0Ddg

For Droserans, Holesky served as a proving ground for our earliest Trappers and Operators. Together, you deployed 33,756 traps and nearly 25,000 nodes. Helping shape a foundation for decentralized automation.

The next step?

Migration to Holesky’s successor: Hoodi.

We’ll be continuing our journey on Hoodi, with even more activities, challenges, and new ways to build using Drosera.

If you want to stay involved, follow the guide below:

https://t.co/hqyAXcoTHr

[Anon Jon @An0n_Jon]

Hop into discord with us and walk through our first set of trap workshops.

[Drosera @DroseraNetwork]

Introducing Trap Academy Workshops

Before TryTrapMe kicks off, we’re hosting a series of hands-on workshops to familiarize the community with traps.

Live on Discord. Recorded for YouTube. Open to all.

📆 Starting Tuesday, May 27
⏰ 10am EST

Day by day, we’ll cover:

1️⃣ What is a trap?
2️⃣ Deploying & running your first trap
3️⃣ Updating Drosera.toml
4️⃣ Editing Trap.sol

By the end of these workshops, you’ll learn how to deploy a trap that immortalizes your Discord username on-chain to earn a special role in the Drosera server!

Sign up below 👇

https://t.co/MS1NcD1gX8