Skip to content
verify mcp Beta VerifyMCP is currently in beta. If you notice any issues, get in touch and we’ll put it right.

Onchain Agent Wallets

NPM · @THREE-WS/ONCHAIN-AGENT-WALLETS

Give an AI agent a Solana spending allowance instead of your key. Chain-enforced cap, x402.

−6 this week 73 Trust /100

Last scanned (13 hours ago)

Scores higher than 46% of npm servers

Category Crypto and Web3

Trust breakdown (7 categories)

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →

Supply Chain Security81
  • No malware found by supply-chain analysis.Pass
  • CVE check failed: a known high-severity CVE affects zod 3.25.76, a direct dependency. A fixed version is available. View diagnostics → Fail
  • No install/post-install scripts declared.Pass
  • 55 of 266 dependencies flagged as unhealthy (1 deprecated). View diagnostics → Partial
Provenance & Transparency48
  • Source repository is publicly reachable at the declared URL. View diagnostics → Pass
  • Provenance check failed: no build-provenance attestation is published. See how to fix → View diagnostics → Fail
  • Clear OSI-approved license (Apache-2.0).Pass
  • Actively maintained (last published 54 days ago).Pass
  • Publishes a security disclosure policy (SECURITY.md).Pass
Schema Quality & AI Usability70
  • AI-judged instruction clarity (excellent).Pass
  • Context-footprint check failed: tool/resource definitions use about 2768 tokens (~197/item across 14 items; 14 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
  • Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management63
  • Stability observed for 19 of 30 days with no destabilising changes; credit accrues until the full window elapses.Partial
Tool Coverage100
  • 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
  • 100% of tool parameters carry a description.Pass
Tool Safety100
  • No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
  • All 6 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation.Pass
  • An AI judge read all 15 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
  • Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.Pass
Changelog

Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.

  • 4 Nov 26 −6
    • Stability: pass → 0.60 functional
  • 29 Oct 26 +1
    • Stability: 0.97 → pass security
  • 28 Oct 26 −1
    • Stability: pass → 0.97 functional
  • 24 Oct 26 −2
    • CVE-2023-54404 affects this package: high ▼ security
    • Known CVEs: CVE check failed: a known high-severity CVE affects zod 3.25.76, a direct dependency. A fixed version is available. security
  • 17 Oct 26 +44
    • Injection markers: unverified → pass ▲ security
    • Stability: unverified → pass ▲ security
    • Tool coverage: unverified → 100 ▲ functional
    • MCP protocol: unverified → pass ▲ functional
    • Dependency health: partial → 0.91 functional
    • Package version: 0.1.0 → 0.1.1 functional
  • 16 Oct 26 +1

    No change was recorded against any check on this day. Supply Chain Security went from 89 to 90.

  • 15 Oct 26 −1

    No change was recorded against any check on this day. Supply Chain Security went from 90 to 89.

  • 14 Oct 26 0
    • Stability: Stability not yet verified: this package could not be installed in a clean environment, so we have no schema to compare. security
    • Tool safety: Tool safety not yet verified: this package could not be installed in a clean environment, so we have no tool text to scan. security
    • Schema quality: Schema quality not yet verified: this package could not be installed in a clean environment, so we have no schema to assess. functional
    • Capabilities: Protocol version not yet verified: this package could not be installed in a clean environment, so we never saw its MCP handshake. functional
    • Tool coverage: Tool coverage not yet verified: this package could not be installed in a clean environment, so we have no tool definitions to assess. functional
    • Dependency health: 0.91 → partial functional
Diagnostics

Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.

Captured 5 Nov 2026 · Analysed npm/@three-ws/onchain-agent-wallets@0.1.1

Provenance No attestation

The registry publishes no build provenance for this version, so there is nothing to verify.

Result No attestation
Ecosystem npm

Background: How many MCP packages publish verified provenance →

Vulnerabilities 6 findings
ID CVE Severity Vector Fix available
GHSA-3gc7-fjrx-p6mg CVE-2025-3194 high CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H no
GHSA-528h-pc64-c93x CVE-2026-71429 medium CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H yes
GHSA-hqr4-qq8f-hg3x CVE-2026-104182 medium CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H yes
GHSA-mjw6-4jj6-33hc CVE-2026-104183 medium CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L yes
GHSA-w5hq-g745-h8pq CVE-2026-41907 high CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N yes
GHSA-mrhx-c7xc-m3w5 CVE-2023-54404 high CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H yes

Background: What a vulnerability scan can and cannot prove →

Dependencies 266 packages
Packages resolved 266
Deprecated 1
Stale 54
No linked repository 1
Tree resolution Complete

Background: SBOMs and build attestations, explained →

MCP tools · 14 exposed · ~2,565 tokens

The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →

Tool Tokens
agent_pay ~154

Send tokens from the agent vault to a recipient, signed by the agent as the on-chain delegate. Every guardrail is checked first (per-transaction cap, rolling daily cap, recipient allowlist, expiry, pause) and the SPL Token program enforces the remaining allowance underneath. A refusal explains which rule stopped it. Requires confirm:true to broadcast.

NameTypeReqDescription
amountstringyesAmount in whole tokens (e.g. "1.25").
confirmboolean–Set true to broadcast.
idstringyesThe agent wallet spending the money.
memostring–Note recorded in the local spend log.
tostringyesRecipient wallet address. Their associated token account is created if needed.

No output schema declared.

No examples provided.

agent_wallet_status ~94

Show an agent wallet as it actually stands on-chain: vault balance, how much of the delegated allowance is left, what it can spend right now, its fee SOL, the guardrails in force, spend totals for the last 24 hours, and any warnings (revoked, expired, paused, out of fee SOL, allowance unbacked by balance). Read-only.

NameTypeReqDescription
idstringyesThe agent wallet.

No output schema declared.

No examples provided.

approve_agent_allowance ~137

Set how much the agent may spend from the vault, enforced by the SPL Token program. This REPLACES the existing allowance, it does not add to it: passing "50" leaves the agent with exactly 50 regardless of what it had before. Use it to raise, lower, or refill an allowance. Requires confirm:true to broadcast.

NameTypeReqDescription
allowancestringyesThe new total allowance in whole tokens. "0" is equivalent to a revoke.
confirmboolean–Set true to broadcast.
idstringyesThe agent wallet.
secretstring–Owner secret key for this call only.

No output schema declared.

No examples provided.

create_agent_wallet ~549

Create an agent wallet on Solana. Generates a fresh keypair for the agent, creates a vault token account that YOU own (derived deterministically from your address plus the agent id), and delegates a capped allowance over it to the agent. The SPL Token program enforces that cap: the agent cannot spend past it, and every spend decrements it. You keep custody, you can revoke in one instruction, and you can withdraw the balance without the agent. Set guardrails (per_tx, daily, allowlists, expiry) in the same call. Signs with OWNER_SECRET_KEY if configured, otherwise returns an unsigned transaction for Phantom, Solflare, Backpack, or a Ledger. Requires confirm:true to broadcast.

NameTypeReqDescription
agent_secretstring–Reuse an existing agent keypair (base58 or JSON array) instead of generating one.
allow_hostsarray–Guardrail: only these hosts may be paid over x402. Subdomains included. Empty means unrestricted.
allow_recipientsarray–Guardrail: only these addresses may receive funds. Empty means unrestricted.
allowancestringyesThe on-chain ceiling, in whole tokens (e.g. "100"). The agent can never spend more than this without a new approval.
confirmboolean–Set true to actually build and broadcast. Without it you get the summary only.
confirm_overstring–Guardrail: spends above this amount need an explicit confirm:true from a human.
dailystring–Guardrail: maximum spend per rolling 24 hours, in tokens.
expires_atstring–Guardrail: ISO timestamp after which every spend is refused.
fee_solstring–SOL to send the agent so it can pay its own transaction fees. Default "0.02". This is the only value the agent itself holds.
idstringyesShort slug naming the agent, lowercase letters/digits/hyphens. Becomes the on-chain vault seed.
labelstring–Human-readable name for your own reference.
mintstring–SPL mint to spend. Defaults to USDC (EPjFWdd5AufqSSqeM2qN1xzybapC8G4wEGGkZwyTDt1v on mainnet).
ownerstring–Your wallet address. Omit to use the configured OWNER_SECRET_KEY. Pass it to keep the key off this machine and sign in your wallet.
per_txstring–Guardrail: maximum single spend, in tokens.
secretstring–Owner secret key for this call only, overriding OWNER_SECRET_KEY.

No output schema declared.

No examples provided.

deploy_agent_onchain ~263

Mint the agent an on-chain identity in the Metaplex Agent Registry on Solana, carrying its EIP-8004 registration document, an x402Support flag, and on-chain attributes naming its payment address, its vault, and the token it accepts. After this the agent is discoverable on metaplex.com/agents and any counterparty can read where to pay it. Costs about 0.007 SOL and needs an owner signing key (set OWNER_SECRET_KEY or pass `secret`); to mint from Phantom or a Ledger instead, use prepare_agent_mint in @three-ws/metaplex-agent-mcp. Requires confirm:true to broadcast.

NameTypeReqDescription
confirmboolean–Set true to broadcast.
descriptionstring–What this agent does.
idstringyesThe agent wallet to publish.
imagestring–Image URL for the asset.
model_urlstring–GLB avatar URL, rendered on metaplex.com/agents.
namestring–On-chain name. Defaults to the agent label or id.
secretstring–Owner secret key for this call only.
servicesarray–Endpoints this agent serves, e.g. its own x402 API.

No output schema declared.

No examples provided.

export_agent_runtime ~139

Produce the config that puts this agent wallet in front of a model: an mcp.json block, the equivalent claude mcp add command, the agent identity card (payment address, vault, token, limits), and the x402 details a counterparty needs to bill it. By default no key material is included, so the output is safe to paste into a repo or a ticket. Pass include_secret:true only when moving the agent to another machine.

NameTypeReqDescription
idstringyesThe agent wallet to export.
include_secretboolean–Include the agent's secret key. Anyone holding it can spend up to the allowance. Default false.

No output schema declared.

No examples provided.

fund_agent_wallet ~148

Move tokens from your wallet into the agent vault, and optionally send the agent a little SOL for its own transaction fees. The vault stays owned by you; funding it does not raise the delegated allowance, so the agent still cannot spend past its ceiling. Requires confirm:true to broadcast.

NameTypeReqDescription
amountstring–Tokens to move into the vault, in whole units (e.g. "50").
confirmboolean–Set true to broadcast.
idstringyesThe agent wallet to fund.
secretstring–Owner secret key for this call only.
solstring–SOL to send the agent for transaction fees (e.g. "0.02").

No output schema declared.

No examples provided.

list_agent_wallets ~50

List the agent wallets configured here, each with its live vault balance, remaining on-chain allowance, and guardrails. Read-only.

NameTypeReqDescription
ownerstring–Only agents belonging to this owner address.

No output schema declared.

No examples provided.

pay_x402 ~223

Call an HTTP endpoint that charges with x402, paying from the agent allowance on Solana. The price is read from the unpaid 402 response first, so the guardrails (per-transaction cap, daily cap, host allowlist, expiry, pause) see the real amount before anything moves. The agent is topped up for exactly that amount from the vault, pays, and is left empty. If the endpoint is not charging, the content comes back with no payment at all. Requires confirm:true when the price is above the confirm threshold or confirmation is on.

NameTypeReqDescription
bodystring–Request body, for POST/PUT/PATCH.
confirmboolean–Set true to authorize payment.
headersobject–Extra request headers.
idstringyesThe agent wallet paying.
max_pricestring–Refuse if the resource asks for more than this, in whole tokens. A ceiling for this call only.
methodstring–HTTP method. Default GET.
urlstringyesThe resource to call.

No output schema declared.

No examples provided.

revoke_agent_wallet ~169

Cancel the agent's delegation on-chain. After this the SPL Token program refuses every transfer the agent signs, permanently, until you approve a new allowance. Pass withdraw:true to also sweep the vault balance back to your wallet in the same transaction. For an instant free stop that needs no transaction and no SOL, use set_guardrails with paused:true instead: that blocks this server, while revoke blocks the chain itself. Requires confirm:true to broadcast.

NameTypeReqDescription
amountstring–Withdraw only this much instead of the whole balance.
confirmboolean–Set true to broadcast.
idstringyesThe agent wallet to revoke.
secretstring–Owner secret key for this call only.
withdrawboolean–Also move the vault balance back to your wallet. Default false.

No output schema declared.

No examples provided.

send_signed_transaction ~118

Broadcast a base64 transaction that Phantom, Solflare, Backpack, or a Ledger already signed, and wait for confirmation. This is how the owner side works without any secret key on this machine: any tool that returns "prepared_for_wallet" hands you a transaction_base64 to sign and pass here. Blockhashes expire in about 60 seconds, so if this reports an expired blockhash, re-run the tool that prepared it.

NameTypeReqDescription
signed_transactionstringyesThe signed transaction, base64-encoded, as your wallet returned it.

No output schema declared.

No examples provided.

set_guardrails ~275

Set the rules this server enforces before it signs a spend: per-transaction cap, rolling 24h cap, allowed recipients, allowed x402 hosts, an expiry, a confirm-above threshold, and a pause switch. Takes effect immediately, costs nothing, and needs no transaction. Only the fields you pass change; pass null to clear one. Empty allowlists mean unrestricted. These sit ON TOP of the on-chain allowance, which no software here can raise: to change that ceiling use approve_agent_allowance, and to cancel it entirely use revoke_agent_wallet.

NameTypeReqDescription
allow_hosts––Only these hosts may be paid over x402, subdomains included. Empty or null means unrestricted.
allow_recipients––Only these addresses may receive funds. Empty or null means unrestricted.
confirm_overstring|null–Spends above this need an explicit human confirm:true. null clears it.
dailystring|null–Maximum spend per rolling 24 hours. null clears it.
expires_atstring|null–ISO timestamp after which every spend is refused. null clears it.
idstringyesThe agent wallet.
pausedboolean–true stops every spend immediately. false resumes.
per_txstring|null–Maximum single spend, in tokens. null clears it.

No output schema declared.

No examples provided.

spend_log ~125

The audit trail for one agent: every spend that went through and every one the guardrails refused, with the amount, destination, transaction signature, and the rule that decided it. Refusals are recorded on purpose, so you can see what an agent tried to do. Read-only.

NameTypeReqDescription
idstringyesThe agent wallet.
limitinteger–How many entries to return, newest first. Default 50.
onlystring–Filter by outcome. Default all.
sincestring–ISO timestamp: only entries at or after this time.

No output schema declared.

No examples provided.

withdraw_from_vault ~121

Withdraw tokens from an agent vault back to your own wallet. The vault is yours, so this always works, with or without a live delegation, and the agent cannot block it. Leaves the allowance untouched: lowering the balance already limits what the agent can spend. Requires confirm:true to broadcast.

NameTypeReqDescription
amountstring–Tokens to withdraw. Omit to withdraw the entire balance.
confirmboolean–Set true to broadcast.
idstringyesThe agent wallet.
secretstring–Owner secret key for this call only.

No output schema declared.

No examples provided.

Install

How do I install the Onchain Agent Wallets MCP server?

Onchain Agent Wallets runs locally as an npm package, launched with npx -y @three-ws/onchain-agent-wallets. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

npm · @three-ws/onchain-agent-wallets

# add to Claude Code
claude mcp add nirholas-onchain-agent-wallets -- npx -y @three-ws/onchain-agent-wallets
Common questions

What is the Onchain Agent Wallets MCP server?

Onchain Agent Wallets is an MCP server listed in the public MCP registry as io.github.nirholas/onchain-agent-wallets. Give an AI agent a Solana spending allowance instead of your key. Chain-enforced cap, x402. This page covers its npm package (@three-ws/onchain-agent-wallets).

Is the Onchain Agent Wallets MCP server safe to use?

Onchain Agent Wallets scores 73 out of 100 on VerifyMCP. We recorded 6 known advisories against it as of 5 November 2026. It declares no install or post-install scripts. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

What tools does the Onchain Agent Wallets MCP server expose?

Onchain Agent Wallets exposes 14 tools: create_agent_wallet, fund_agent_wallet, approve_agent_allowance, set_guardrails, agent_pay, and 9 more. Their descriptions and schemas cost roughly 2,565 tokens of context every time the server is loaded.

Is the Onchain Agent Wallets MCP server still maintained?

Onchain Agent Wallets is still listed as active in the MCP registry. We last reached this channel on 5 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

What licence is the Onchain Agent Wallets MCP server under?

Onchain Agent Wallets declares the Apache-2.0 licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.