HelmProtocol

Helm is the safety layer for autonomous agents.
Programmable multisig on @solana

[HelmProtocol @Helmprotocol]

Helm is now one npm install away.

@helmmasters/agentvault-sdk v0.1.0 is live. TypeScript builders for every Helm program instruction, account decoders, PDA helpers — everything you need to plug an agent into the trust layer.

npm install @helmmasters/agentvault-sdk

If you're building an agent that touches money on Solana, the integration is short:
1. Create a vault, set caps + allowlist
2. Give your agent its keypair
3. Call proposeTransactionIx whenever it wants to act
4. Helm enforces the policy on-chain — instant tier executes, delayed tier waits for your sign-off, over-cap rejects

The reference agent (https://t.co/1xViN988Bc → examples/mainnet-agent) shows the propose flow end to end on mainnet.

Apache-2.0. Mainnet PROGRAM_ID baked in. Custom deployments via programId override.

The trust layer is here. Plug your agent in.

→ https://t.co/A4zM1Z2Ypd
→ https://t.co/1xViN988Bc
→ https://t.co/mYUyn4wrc9

[HelmProtocol @Helmprotocol]

@bankrbot
Launch this on Robinhood chain name it Agent Multisig, Ticker HELM

[HelmProtocol @Helmprotocol]

Agentic API economy testnet on Solana
Now moving to RobinHood chain for Mainnet

[HelmProtocol @Helmprotocol]

@MasterAlpha27 We will migrate to RH Agentic area.

[HelmProtocol @Helmprotocol]

Agentic API economy testnet on Solana
Now moving to RobinHood chain for Mainnet

[HelmProtocol @Helmprotocol]

@coracarter3127 @NousResearch Putting limit on your Agent edges

[HelmProtocol @Helmprotocol]

@helmmasters/agentvault-mcp is live.

Any agent that speaks MCP — @NousResearch -Hermes, Claude Desktop, Cursor, anything else in the ecosystem — can now manage Helm vaults directly: create them, set policy, propose payments, approve or veto, all through a clean MCP interface.

17 tools. Read-only for unsigned setups, full write access when you point it at a keypair. SDK underneath is the same one already on npm. No new trust assumptions.

npm install -g @helmmasters/agentvault-mcp

→ https://t.co/fMC31C2U3G

[HelmProtocol @Helmprotocol]

A note on agents and money.

The interesting agents right now — the ones that grow, persist across sessions, live in your Telegram or Discord, manage projects over weeks — none of them can spend money. They reason, they remember, they delegate, they automate. But economic action? Still you, manually.

The bottleneck isn't the agent. It's the trust model. Wiring an LLM to a hot wallet is reckless. Wiring it to a corporate-controlled payment rail kills the openness.

There's a third way that's been quietly possible for a few months: programmable on-chain vaults that let an agent spend, but only within rules you set. Caps in USD, allowlists by program, guardian veto. The agent operates; the chain enforces; you sleep.

Helm is that layer. SDK is on npm. Mainnet program is live. The dashboard runs in your browser. The Telegram bot notifies you when something needs your sign-off.

What's missing is the last mile — making it trivially callable by the agents that already exist. That's the next thing we're shipping.

If you're building or running a persistent agent and the "give it money to act on your behalf" question keeps coming up, DMs open.

→ https://t.co/A4zM1Z2Ypd

[HelmProtocol @Helmprotocol]

A note on agents and money.

The interesting agents right now — the ones that grow, persist across sessions, live in your Telegram or Discord, manage projects over weeks — none of them can spend money. They reason, they remember, they delegate, they automate. But economic action? Still you, manually.

The bottleneck isn't the agent. It's the trust model. Wiring an LLM to a hot wallet is reckless. Wiring it to a corporate-controlled payment rail kills the openness.

There's a third way that's been quietly possible for a few months: programmable on-chain vaults that let an agent spend, but only within rules you set. Caps in USD, allowlists by program, guardian veto. The agent operates; the chain enforces; you sleep.

Helm is that layer. SDK is on npm. Mainnet program is live. The dashboard runs in your browser. The Telegram bot notifies you when something needs your sign-off.

What's missing is the last mile — making it trivially callable by the agents that already exist. That's the next thing we're shipping.

If you're building or running a persistent agent and the "give it money to act on your behalf" question keeps coming up, DMs open.

→ https://t.co/A4zM1Z2Ypd

[HelmProtocol @Helmprotocol]

Yesterday I scheduled a policy update on my mainnet vault. SPL Token to the whitelist. USDC mint to the token allowlist. The price feed switched from SOL/USD to USDC/USD so the cap math actually works for stablecoin payments.

The on-chain timelock is 24 hours.(Helm SafeLayer Policy) So I'm sitting here, watching my own protocol make me wait. Funded the vault with $15 USDC in the meantime.

TX https://t.co/EYKvhQsg6T

While the clock runs, the rest of the stack came together:
→ Local x402 server with two paid endpoints ($0.10 and $10)
→ Agent loop that hits the API, parses the 402 challenge, proposes the payment through the vault, polls for guardian approval, executes
→ End-to-end dry run passed — agent talks to live server, builds the propose ix correctly, derives the right PDAs

Tomorrow, when the timelock clears:
→ Apply the policy update
→ Run the agent against the real server
→ $0.10 payment lands as a pending tx → I approve from Telegram → vault settles → API returns 200
→ $10 payment goes through the same flow

Every payment routed by an agent. Every payment gated by the vault. Every cap enforced on Solana. The agent never touches your wallet, only proposes against your policy.

Pre-flight is clean.

[HelmProtocol @Helmprotocol]

Yesterday I scheduled a policy update on my mainnet vault. SPL Token to the whitelist. USDC mint to the token allowlist. The price feed switched from SOL/USD to USDC/USD so the cap math actually works for stablecoin payments.

The on-chain timelock is 24 hours.(Helm SafeLayer Policy) So I'm sitting here, watching my own protocol make me wait. Funded the vault with $15 USDC in the meantime.

TX https://t.co/EYKvhQsg6T

While the clock runs, the rest of the stack came together:
→ Local x402 server with two paid endpoints ($0.10 and $10)
→ Agent loop that hits the API, parses the 402 challenge, proposes the payment through the vault, polls for guardian approval, executes
→ End-to-end dry run passed — agent talks to live server, builds the propose ix correctly, derives the right PDAs

Tomorrow, when the timelock clears:
→ Apply the policy update
→ Run the agent against the real server
→ $0.10 payment lands as a pending tx → I approve from Telegram → vault settles → API returns 200
→ $10 payment goes through the same flow

Every payment routed by an agent. Every payment gated by the vault. Every cap enforced on Solana. The agent never touches your wallet, only proposes against your policy.

Pre-flight is clean.

[HelmProtocol @Helmprotocol]

-Helm sdk Audit.

https://t.co/AcsPGmr2aX

[HelmProtocol @Helmprotocol]

Time to Building: x402 + Helm.

Solana is settling the agentic API economy. Agents pay for endpoints, no accounts, no API keys — just micropayments.

Question that hasn't been answered yet: who enforces the spend limit?

Today, the agent's wallet does. Whatever's in it is at risk if the agent gets prompt-injected, key-leaked, or simply behaves unexpectedly.

We're wiring x402 through Helm. Same flow — agent hits a paid API, gets a 402, signs a payment — except the payment goes through a programmable vault. Small payments execute instantly under the cap. Large ones land in your Telegram for approval. The vault enforces. The chain settles. The agent doesn't carry the blast radius.

Helm Agents will able use real USDC, real x402 endpoint, real Helm vault.

If you're building an x402 facilitator, API server, or paying agent — and want to chat about the integration shape —

→ https://t.co/A4zM1Z2Ypd

[HelmProtocol @Helmprotocol]

Helm is now one npm install away.

@helmmasters/agentvault-sdk v0.1.0 is live. TypeScript builders for every Helm program instruction, account decoders, PDA helpers — everything you need to plug an agent into the trust layer.

npm install @helmmasters/agentvault-sdk

If you're building an agent that touches money on Solana, the integration is short:
1. Create a vault, set caps + allowlist
2. Give your agent its keypair
3. Call proposeTransactionIx whenever it wants to act
4. Helm enforces the policy on-chain — instant tier executes, delayed tier waits for your sign-off, over-cap rejects

The reference agent (https://t.co/1xViN988Bc → examples/mainnet-agent) shows the propose flow end to end on mainnet.

Apache-2.0. Mainnet PROGRAM_ID baked in. Custom deployments via programId override.

The trust layer is here. Plug your agent in.

→ https://t.co/A4zM1Z2Ypd
→ https://t.co/1xViN988Bc
→ https://t.co/mYUyn4wrc9

[HelmProtocol @Helmprotocol]

@RetardedNi85688 Ledger out
Helm In

[HelmProtocol @Helmprotocol]

@King_Memento https://t.co/x5HSeZ9fZn

[HelmProtocol @Helmprotocol]

Helm is now one npm install away.

@helmmasters/agentvault-sdk v0.1.0 is live. TypeScript builders for every Helm program instruction, account decoders, PDA helpers — everything you need to plug an agent into the trust layer.

npm install @helmmasters/agentvault-sdk

If you're building an agent that touches money on Solana, the integration is short:
1. Create a vault, set caps + allowlist
2. Give your agent its keypair
3. Call proposeTransactionIx whenever it wants to act
4. Helm enforces the policy on-chain — instant tier executes, delayed tier waits for your sign-off, over-cap rejects

The reference agent (https://t.co/1xViN988Bc → examples/mainnet-agent) shows the propose flow end to end on mainnet.

Apache-2.0. Mainnet PROGRAM_ID baked in. Custom deployments via programId override.

The trust layer is here. Plug your agent in.

→ https://t.co/A4zM1Z2Ypd
→ https://t.co/1xViN988Bc
→ https://t.co/mYUyn4wrc9

[HelmProtocol @Helmprotocol]

@King_Memento We guess you're not aware of the updates we make on GitHub every day.

[HelmProtocol @Helmprotocol]

@Aminu1281578 yes

[HelmProtocol @Helmprotocol]

Helm now has an official Telegram channel.

Product updates, release notes, security disclosures, and the occasional thinking on agent infrastructure — without the price talk, the hype cycles, or the daily noise.

→ https://t.co/6GXhkaBMha

If you've connected @ HelmProtocolBot to a vault, you're already getting real-time pings on pending transactions. The channel is for everything else — what's shipping, what's coming, what we're learning.

Building the trust layer for AI agents on Solana. In public.

[HelmProtocol @Helmprotocol]

A note for builders shipping agents on Solana:

If your agent has a wallet, you're already designing its trust model — whether you've thought about it or not. The question is whether that model lives in your code (where any bug, prompt injection, or compromised key drains everything) or on-chain (where the protocol enforces the rules and the worst-case is bounded).

Helm is the second option, productized.

You give your agent a vault instead of a raw wallet. You set USD spending caps. You set the program allowlist. You hold veto rights on anything above the instant tier. Pyth prices every transaction live. Solana enforces every constraint.

The agent's code stays simple — it just proposes transactions. The hard parts (price feeds, policy enforcement, guardian approval flow, timelocked policy updates) all happen in the protocol.

Integration is one npm install away — coming this week. Until then:

→ The protocol is live on mainnet
→ The dashboard works (https://t.co/Mt0H8K7XGV)
→ The Telegram bot notifies guardians in real time
→ The reference agent (examples/mainnet-agent) shows the propose flow end to end

If you're building an agent that touches money, the integration shape is:
1. Create a vault, set caps + allowlist
2. Give your agent the agent keypair
3. Call propose_transaction whenever the agent wants to act
4. Helm handles the rest

https://t.co/1xViN988Bc

The trust layer is here. Plug your agent in.

[HelmProtocol @Helmprotocol]

@iinvestoradesam We will release an SDK before today

[HelmProtocol @Helmprotocol]

A note for builders shipping agents on Solana:

If your agent has a wallet, you're already designing its trust model — whether you've thought about it or not. The question is whether that model lives in your code (where any bug, prompt injection, or compromised key drains everything) or on-chain (where the protocol enforces the rules and the worst-case is bounded).

Helm is the second option, productized.

You give your agent a vault instead of a raw wallet. You set USD spending caps. You set the program allowlist. You hold veto rights on anything above the instant tier. Pyth prices every transaction live. Solana enforces every constraint.

The agent's code stays simple — it just proposes transactions. The hard parts (price feeds, policy enforcement, guardian approval flow, timelocked policy updates) all happen in the protocol.

Integration is one npm install away — coming this week. Until then:

→ The protocol is live on mainnet
→ The dashboard works (https://t.co/Mt0H8K7XGV)
→ The Telegram bot notifies guardians in real time
→ The reference agent (examples/mainnet-agent) shows the propose flow end to end

If you're building an agent that touches money, the integration shape is:
1. Create a vault, set caps + allowlist
2. Give your agent the agent keypair
3. Call propose_transaction whenever the agent wants to act
4. Helm handles the rest

https://t.co/1xViN988Bc

The trust layer is here. Plug your agent in.

[HelmProtocol @Helmprotocol]

@faizantariq26 Shout out to @Helius

[HelmProtocol @Helmprotocol]

Just watched my agent on Solana mainnet propose a $10 transfer.

Two seconds later, Telegram buzzed:

🔔 New pending transaction
Vault: AQt7…8cPS
Action: 0.14 SOL → unknown program
USD: $10.00 (hourly: $0 / $25)

[Approve] [Veto] [View vault]

Tapped Approve → https://t.co/Mt0H8K7XGV opened → Phantom signed → done. The agent's transaction cleared on-chain.

Or I could've tapped Veto. Or just ignored it for 24 hours and let the timelock expire.

This is the agent economy with a kill switch in your pocket.

The bot never touched my keys. It just told me what my agent wanted to do, and routed me to my own wallet to make the call.

Connect your own at https://t.co/Mt0H8K7XGV → vault detail → Connect Telegram.

@HelmProtocolBot

[HelmProtocol @Helmprotocol]

You can try now.

[HelmProtocol @Helmprotocol]

Real scenarios this unlocks:

→ Shopping agent with a Solana wallet — let it browse and buy, but every purchase pings your phone for approval before it clears
→ Polymarket trading agent — set a $50/tx cap, then catch any large position your agent wants to open before it commits
→ Treasury automation — your DAO agent rebalances daily, but anything above the daily cap waits for a human in the chat
→ Subscription / payments agent — small recurring payments flow through, but new merchants always trigger a delayed approval
→ Yield farming agent — auto-compound under the cap, ask permission before moving size

The pattern is the same every time: agent acts within the rules instantly, asks you for anything outside them. The chain enforces. The chat notifies. You sign or veto on your own terms.