@leashmarket/mcp

Standalone MCP server for the Leash identity layer for AI agents. It lets any AI agent in any MCP host (Cursor, Claude Desktop, Cline, Continue, ChatGPT-MCP, …) resolve and verify identities, inspect proof trails, sign on-chain Solana transactions, pay x402 paywalls, and MPP paywalls, create hosted payment links, and check its treasury balance — without a browser in the loop.

Install

// In Cursor → Settings → MCP, or your MCP host's equivalent:
{
  "mcpServers": {
    "leash": {
      "command": "npx",
      "args": ["-y", "@leashmarket/mcp"],
    },
  },
}

Most MCP hosts support an env map on each server — use it to override the default public RPC (slow / rate-limited). Swap in your own URL:

{
  "mcpServers": {
    "leash": {
      "command": "npx",
      "args": ["-y", "@leashmarket/mcp"],
      "env": {
        "LEASH_RPC_URL": "https://devnet.helius-rpc.com/?api-key=YOUR_KEY",
        // optional — match mainnet if your agent + links are on mainnet:
        // "LEASH_NETWORK": "solana-mainnet",
        // "LEASH_RPC_URL": "https://mainnet.helius-rpc.com/?api-key=YOUR_KEY",
      },
    },
  },
}

You can set any other overrides the same way (LEASH_AGENT_MINT, LEASH_EXECUTIVE_KEY, LEASH_API_URL, LEASH_EXPLORER_URL, …). Alternatively, put rpc_url in ~/.config/leash/agent.json — env wins over the file when both are set.

Configure

The server looks at, in order:

1. ~/.config/leash/agent.json (chmod 600). Same posture as gcloud/gh/aws.

2. Environment variables (override the file when set):

   | env | required | example | | --------------------- | ------------------------ | --------------------------------------------- | | LEASH_AGENT_MINT | yes | Agnt7XQ... | | LEASH_EXECUTIVE_KEY | yes | 5Jz... (base58) or [12,34,...] (JSON arr) | | LEASH_NETWORK | no | solana-mainnet (default) / solana-devnet | | LEASH_API_URL | no | https://api.leash.market (default) | | LEASH_RPC_URL | strongly recommended | bring your own — see below | | LEASH_EXPLORER_URL | no | https://explorer.leash.market (default) | | LEASH_API_KEY | no | legacy bearer for /v1/payment-links | | LEASH_PER_CALL_USDC | no | per-call spend cap (default 1) | | LEASH_PER_DAY_USDC | no | per-day spend cap (default 10) |

Bring your own RPC. The default endpoints (api.devnet.solana.com, api.mainnet-beta.solana.com) are public, rate-limited, and slow. Each leash_pay_payment_link makes 3-5 RPC calls — on a public endpoint that's a 4-8s settlement, sometimes a 429. Set LEASH_RPC_URL (or rpc_url in agent.json) to a Helius / Triton / QuickNode / Alchemy / self-hosted endpoint and settlement drops under one second.

The server starts without an agent configured — tools/list still works, but every tool short-circuits with a no_agent JSON blob asking the LLM to onboard the user. (The frictionless leash_register_agent tool ships in the next release.)

agent.json example:

{
  "version": 1,
  "agent_mint": "Agnt7XQ...",
  "executive_keypair": "5Jz...", // base58 OR a 64-element JSON array
  "network": "solana-devnet",
  "rpc_url": "https://devnet.helius-rpc.com/?api-key=YOUR_KEY",
  "explorer_url": "https://explorer.leash.market",
  "created_at": "2026-04-30T...",
}

Tools (28 canonical)

| Tool | What it does | | ---------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | leash_register_agent | Two-step onboarding (call this tool TWICE). Step 1: collect agent name + description + image_url + EIP-8004 services[] from the user, persist them alongside a generated/imported executive keypair, return funding_required. Step 2: after the user funds the executive with SOL, mint MPL Core agent, set unlimited USDC delegation, record on the API, and HOT-SWAP the in-memory MCP host so subsequent tool calls work without a restart. | | leash_get_identity | Self-introspection — what agent am I, on which network, what's my treasury PDA. | | leash_resolve_identity | Resolve another agent by mint, handle, or verified domain. Returns public profile, verified domains, public capability cards, public claims, and reputation summary. | | leash_verify_identity | Verify that a mint, handle, or domain resolves to a live Leash identity. Add intent/capability/thresholds to get an allow/warn/deny trust verdict before paying, trusting a claim, or calling a capability. | | leash_get_identity_profile | Fetch the active agent's editable profile, including private capability cards and active claims. | | leash_update_identity_profile | Update the active agent handle, visibility metadata, and full capability-card array via X-Leash-Sig. | | leash_verify_identity_domain | Verify a domain selector for the active agent via /.well-known/leash-agent.json. | | leash_create_identity_claim | Attach a signed public/private claim to the active agent identity. | | leash_revoke_identity_claim | Revoke one claim owned by the active agent. Revoked claims stop resolving publicly and stop reading through disclosures. | | leash_list_identity_disclosures | List selective-disclosure grants created by the active agent. Tokens are not re-shown. | | leash_create_identity_disclosure | Create a bearer-token disclosure grant for private capability cards, private claims, or selected receipt fields. | | leash_revoke_identity_disclosure | Revoke one disclosure grant owned by the active agent. | | leash_check_treasury_balance | Read SOL + USDC/USDG/USDT balances on the agent treasury PDA. | | leash_create_agent_api_key | Create an agent scoped API key for the active agent by signing with the local executive keypair. Plaintext is returned once. | | leash_list_agent_api_keys | List the active agent's API keys. Returns id, label, prefix, last4, scope, and timestamps; never returns plaintext. | | leash_revoke_agent_api_key | Disable one active-agent API key by id. Enforcement is by signed agent mint, so one agent cannot revoke another agent's key. | | leash_pay_payment_link | Probe an x402/MPP link, sign + settle the SPL transfer locally, and return the receipt. Accepts method + body for POST paywalls. | | leash_create_payment_link | Mint an x402/MPP paywall the user can share. Set upstream_url to monetize an existing API and expected_request_body to document POST body metadata. Uses LEASH_API_KEY; create one first with leash_create_agent_api_key if needed. | | leash_withdraw_treasury | Owner-driven withdrawal of SOL or an SPL stable to any wallet (mpl-core::Execute). | | leash_set_spend_limit | Owner-driven update of the SPL Approve delegation that lets the executive spend stables from the treasury. mode: 'unlimited' \| 'revoke' \| 'amount' — tighten, pause, or restore the cap. | | leash_get_spend_limit | Read the current SPL delegation + treasury balance for a stable. Reports delegate, remaining cap (atomic + decimal), and balance. | | leash_receipts | Paginated receipts feed for the active agent. Requires LEASH_API_KEY until X-Leash-Sig auth lands. | | leash_get_receipt | Fetch a single ReceiptV1 by receipt_hash — same canonical JSON the explorer renders at /receipt/{hash} (full price legs, request URL, decision, tx_sig, prev/current chain). | | leash_transaction_history | List every earn + spend receipt in the last N days (default 7) plus running USD totals (total_sent_usd, total_received_usd, net_usd). Stables (USDC/USDG/USDT) summed at 1:1. | | leash_daily_transactions | Bin the same window into per-day buckets [{ date, sent_usd, received_usd, net_usd, sent_count, received_count }] plus grand totals — structured P&L view. | | leash_discover | Search the Leash marketplace and pay.sh/pay-skills registry for paid services by capability + price. Public read — works without an agent. | | leash_pay_skills_endpoints | Expand a pay.sh/pay-skills provider into payable endpoint URLs. | | leash_reputation | Live reputation snapshot for any on-chain agent — settled-call volume, dispute rate, distinct counterparties. Public read. |

Subcommands (cross-interface portability)

The CLI is more than the STDIO server — leash-mcp has a small set of subcommands so an agent can move freely between hosts:

leash-mcp                  # default — run the STDIO MCP server
leash-mcp export           # print active agent.json to stdout
leash-mcp export --out a.json   # save instead
leash-mcp import path/to/agent.json  # install into ~/.config/leash/
leash-mcp doctor           # config + RPC + API reachability check
leash-mcp help             # full help

Use export + import to roam: an agent minted from Cursor's MCP can be exported, dropped into Claude Desktop, and imported into its config — same on-chain identity, same treasury, same reputation. Same JSON also pastes cleanly into the chat product's Profile → Agent → Import page (forthcoming).

Hosted paywalls for existing APIs

leash_create_payment_link can create a hosted /x/{id} URL in front of an API you already run:

{
  "label": "Design agent",
  "amount": 1,
  "currency": "USDC",
  "method": "POST",
  "protocol": "x402",
  "upstream_url": "https://api.example.com/design",
  "expected_request_body": {
    "prompt": "string",
    "style": "string",
    "format": "string"
  }
}

expected_request_body is discovery metadata only. The buyer agent sends the real JSON body later through leash_pay_payment_link:

{
  "url": "https://api.leash.market/x/design-agent?network=solana-devnet",
  "method": "POST",
  "body": "{\"prompt\":\"Design a landing page\",\"style\":\"premium dark mode\"}"
}

After settlement, Leash forwards that buyer body to upstream_url and returns the upstream response.

Use protocol: "mpp" when the hosted paywall should speak the MPP problem+json flow instead of x402's HTTP 402 flow.

Try the read path

After provisioning an agent with leash_register_agent (or leash agent create), poke balances through the MCP protocol directly:

LEASH_AGENT_MINT=<mint> \
LEASH_EXECUTIVE_KEY=<base58 secret> \
LEASH_NETWORK=solana-mainnet \
pnpm --filter @leashmarket/mcp dev:demo-balance

That bypasses STDIO and uses an in-memory transport — fastest way to verify the path before recording a real demo.

Develop

pnpm --filter @leashmarket/mcp typecheck
pnpm --filter @leashmarket/mcp test
pnpm --filter @leashmarket/mcp build

The compiled dist/cli.js is set executable (chmod +x) by the build script so npx -y @leashmarket/mcp works without an extra step.