The universal execution layer. Deploy once and reach every chain. Explore: portal.push.org
Global
- Tweets 10,473
- Following 185
- Followers 158,408
- Likes 20,364
Pinned Tweet
Push Chain has successfully completed its security audit with @hackenclub.
By the numbers:
- Zero critical-severity findings were identified
- 300 hours of audit work
- 3 smart contracts and full chain audit completed
- $100,000 bug bounty launching soon
62
67
262
62,746
Cheap work can beat good work if both get paid the same.
Suppose an agentic job costs $1.
Producing a careful output costs the provider $0.20 in model inference, data calls, and computation. A fast low-quality answer costs only $0.03.
If both outputs require the same amount of payment, the economic incentive is clear.
6
4
32
3,689
You cannot ignore the effort Solana is putting into becoming the leading settlement layer for agentic transactions.
There’s no clearer demand signal for any emerging tech than industry leaders going all in.
☑ 6 months at @SolanaFndn
→ reporting for duty.
My biggest focus has been on agentic payments: making sure @solana is the perfect fit for this.
🧵 Unbundling our current stack
2
24
3,544
Push Pulse 🫀 - Pulse 48
→ shipped: things we can’t show you yet
→ merged: opinions
→ reverted: one opinion
→ nomenclature debates: 1 closed, 3 opened
So here’s the update: we're cooking something in stealth.
10
3
29
4,224
Let me repeat:
"The number of agents will keep growing"
INSIGHT: @Coinbase CEO @brian_armstrong says "The number of agents will keep growing. Crypto and stablecoins will be their go-to payment rail."
2
3
20
4,261
𝟭 𝗠𝗶𝗻𝘂𝘁𝗲 𝘄𝗶𝘁𝗵 𝗣𝘂𝘀𝗵 ⚡ — 𝗪𝗲𝗲𝗸 𝟰𝟴
Heads-down build week. Here's where things stand 👇
🔧 𝗥𝗲𝗮𝗱-𝘀𝘁𝗮𝘁𝗲 𝗦𝗗𝗞
We're wrapping up the SDK that lets apps read state directly.
✅ 𝗘𝗥𝗖-𝟴𝟭𝟴𝟯
The first implementation is done, with hooks.
- Every Job ID is now tied to a Mandate ID
- One AGW, one job, enforced right in the hook
Next up: letting existing mandates be modified.
🔬 𝗥&𝗗: 𝗔𝗴𝗲𝗻𝘁 𝗝𝗼𝗯 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲
Two design proposals are being explored side by side. We're also building more hooks for different job types.
⏭️ 𝗔𝗚𝗪 𝗦𝗗𝗞
Kicks off this weekend or early next week.
𝗪𝗲𝗲𝗸𝗲𝗻𝗱 𝗩𝗶𝗯𝗲𝘀 🍩 → agents don't take weekends off, but the humans are trying.
3
1
28
3,246
𝗣𝘂𝘀𝗵 𝗣𝘂𝗹𝘀𝗲 🫀 — 𝗣𝘂𝗹𝘀𝗲 𝟰𝟳
Read State is live in the Docs Playground ⚡
One batch. Three reads:
→ Ethereum price feed
→ Solana token balance
→ Public API
Hit run ▶️ push.org/docs/chain/build/re…
4
3
21
3,116
WE ARE ABOUT TO WITNESS AN ERA OF HYPERPARABOLIC AGENTIC ACTIVITY ONCHAIN.
THE SIGNS ARE VISIBLE.
GET READY.
Our latest research paper explores the growing connection between AI and digital assets and explains why broad AI adoption may drive new demand, utility and applications across the digital asset economy. blackrock.com/us/individual/…
17
3
29
3,898
If you think AI agents are as loyal as a golden retriever, you need to read this ↓↓
Disturbing findings about AI agents you cannot ignore:
1. Out of 700, 87% of big scale orgs reported more than one security incidents with their deployed agents.
2. Google saw a 32% increase in malicious indirect prompt-injection detections btwn nov’25 - feb’26
3. 82% of Enterprises Have Unknown AI Agents in Their Environments
4. 3,250 attack scenarios across seven models with a 73.5% success rate
3
2
23
5,122
𝟭 𝗠𝗶𝗻𝘂𝘁𝗲 𝘄𝗶𝘁𝗵 𝗣𝘂𝘀𝗵 ⚡ — 𝗪𝗲𝗲𝗸 𝟰𝟳
📖 𝗥𝗲𝗮𝗱 𝗦𝘁𝗮𝘁𝗲
Live on testnet. A contract on Push Chain can now read state from a contract on any supported chain.
🧩 𝗪𝗮𝗹𝗹𝗲𝘁 & 𝗦𝗲𝘁𝘁𝗹𝗲𝗺𝗲𝗻𝘁
- Wallet v1 architecture is finalized.
- SDK work has started on the feature set that talks to it.
🔐 𝗔𝘂𝗱𝗶𝘁
Bug bounty is complete and closed.
🧱 𝗡𝗲𝘅𝘁
Preliminary architecture for 8183 is finalized.
𝗪𝗲𝗲𝗸𝗲𝗻𝗱 𝗩𝗶𝗯𝗲𝘀 🍩: reading state, not the room.
8
3
29
3,490
TL;DR:
1. Give every agent its own identity, wallet, and audit trail.
2. Treat every website, API, and agent message as hostile input.
3. Lock memory writes with source, trust level, and expiry.
4. Always assume the agent will fail at some point so place effective measures to contain the blast radius.
Read rest of the tips here:
3
2
19
3,202
POV:
a malicious prompt bear-hugging my perfectly innocent agent to trick it into draining its wallet.
3
2
17
3,196
Disturbing findings about AI agents you cannot ignore:
1. Out of 700, 87% of big scale orgs reported more than one security incidents with their deployed agents.
2. Google saw a 32% increase in malicious indirect prompt-injection detections btwn nov’25 - feb’26
3. 82% of Enterprises Have Unknown AI Agents in Their Environments
4. 3,250 attack scenarios across seven models with a 73.5% success rate
2
3
24
7,808
sources:
1. harness.io/state-of-agent-dl…
2. cloudsecurityalliance.org/pr…
3. blog.google/security/prompt-…
4.sciencedirect.com/science/ar…
2
424