Skip to main content
Glama

consensa-mcp

An MCP server that lets an AI agent record a consented, attributed, paid-for receipt for the open-source dependencies it consumes — settled per call in USDC on Algorand via Consensa. Each call splits payment on-chain: provider 80%, upstream maintainers' escrow 11% (per-dependency attribution), commons 7%, protocol 2%.

Tools

  • consensa_attribution_preview(manifest) — free. Which upstream deps would be attributed on-chain, plus the split and price. No payment.

  • consensa_clearance(manifest)paid. Pays USDC from the configured wallet and returns the on-chain receipt (paymentTxId, settleTxId, split, upstream).

  • consensa_receipt(paymentTxId) — free. Look up a prior receipt.

Related MCP server: @true402.dev/mcp-server

Bring-your-own-wallet

You run this server with your own funded Algorand spend wallet (CONSENSA_MCP_PAYER_MNEMONIC). Consensa never holds or spends your funds. Use a dedicated, low-balance wallet.

Try it free — no wallet, no keys, right now

Two of the three tools are plain HTTP against the live endpoint. They need no mnemonic, no funded account, and no configuration beyond the defaults:

  • consensa_attribution_preview(manifest) — hand it a package.json and see exactly which upstream packages a paid call would attribute on-chain, the four-way split, and the price. Nothing is paid and nothing is written to any chain.

  • consensa_receipt(paymentTxId) — look up any receipt this endpoint has issued.

Run the server with no env at all and ask your agent to preview a manifest. This is the honest "$0 trial": the preview is the same computation the paid call performs, from the same endpoint, over the same terms — it simply stops before paying.

Configure

Copy .env.example and fill it in — see that file for every variable. The payer mnemonic must come from your environment; never commit it.

CONSENSA_NETWORK has no default and must match the endpoint you point at. The server asks the endpoint which network it serves and refuses on conflict rather than guessing. Against a mainnet endpoint it also refuses to pay until you have written mainnet yourself — spending real USDC should be something you said, not something a default decided.

Spend guards (defaults): MAX_SPEND_USDC_PER_RUN=5.00, MAX_PER_CALL_USDC=0.02, MAX_CALLS_PER_MIN=10, plus idempotency (identical manifest within the TTL is not charged twice). Optionally pin CONSENSA_EXPECTED_PAYTO so the server refuses to pay any other address.

Run (from the cloned repo)

npm install
npm run dev          # tsx src/index.ts   (or: npm run build && npm start)

The server speaks MCP over stdio. With no env set it serves the two free tools. For the paid tool, add the wallet and network:

$env:CONSENSA_NETWORK="mainnet"; $env:CONSENSA_MCP_PAYER_MNEMONIC="<25 words>"

Use it on Mainnet — from an MCP client (e.g. Claude Desktop)

{
  "mcpServers": {
    "consensa": {
      "command": "npx",
      "args": ["tsx", "src/index.ts"],
      "cwd": "/absolute/path/to/consensa-mcp",
      "env": {
        "CONSENSA_NETWORK": "mainnet",
        "CONSENSA_ENDPOINT": "https://consensa-endpoint-production.up.railway.app",
        "CONSENSA_EXPECTED_PAYTO": "PJOXFIBKVKGYGMCW3346P5N4OATMODZYM3VQOXDAEUTIGAOGRYGTIO3NME",
        "CONSENSA_MCP_PAYER_MNEMONIC": "<your 25-word mnemonic>",
        "MAX_SPEND_USDC_PER_RUN": "1.00"
      }
    }
  }
}

Status

v0.2, validated end-to-end on Algorand MainNet: payment BADUNG7YO263XRACSVHASSRJ5MIHXQRKDKUYKIRU42BAQEQCRXMA (10,000 microUSDC) settled by GYVS26OZABS4E6FQEH5C5YQBIV5ULG4VBMBVOYXCX4HG3FJ7UWGA, whose on-chain receipt records the split as 8,000 / 1,100 / 700 / 200 — summing exactly to the payment. Bring-your-own-wallet. Start with the free tools above — they cost nothing and show you exactly what a paid call would record.

A
license - permissive license
Not graded
quality - not tested
C
maintenance

Maintenance

Maintainers
Response time
Release cycle
Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Servers

View all related MCP servers

Related MCP Connectors

View all MCP Connectors

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/consensa-algo/consensa-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server