Sparring
Launch app
Early access

Run your onchain agent against a pack of traps (fake invoices, inflated 402 prices, poisoned token names) before it holds real funds.

The scenarios are specific to agents that move money on Solana and pay over x402. Generic LLM red-team tools test what a model says, while Sparring tests what the agent does with a wallet in hand.

Launch app Read the docs
Sparring
  • Mint authority revokedpass
  • Freeze authority revokedpass
  • Top 10 hold 39.0%pass
  • Liquidity lockedcheck
  • Metadata immutablepass

Score

66 / 100

Recent checks

  • KALI2 h ago
  • DOMO2 h ago
  • NEQ2 h ago
Connect an agent by HTTP endpoint or MCP server, with a mock wallet and mock RPC tools so no real funds are involvedStarter pack of 30 scenarios: prompt injection through token metadata, address swap in an invoice, inflated or repeatedPer-scenario pass or fail with a replayable transcript showing exactly what the agent decided and whyShareable report page and badge showing which scenarios the agent passed and the date of the runGitHub Action that reruns the pack on every release and fails the build on a regression
Features

What you get with Sparring.

Built for teams shipping trading, payment or procurement agents, solo builders about to fund an agent for the first time, and funds that want evidence before they.

Share

One link with the full report.

Connect an agent by HTTP endpoint or MCP server, with a mock wallet and mock RPC tools so no real funds are involved

Planned

Starter pack of 30 scenarios: prompt injection through token metadata, address swap in an invoice, inflated or repeated

Planned

Per-scenario pass or fail with a replayable transcript showing exactly what the agent decided and why

Planned

Shareable report page and badge showing which scenarios the agent passed and the date of the run

Planned

GitHub Action that reruns the pack on every release and fails the build on a regression

Planned
Report

Connect an agent by HTTP endpoint or MCP server, with a mock wallet and mock RPC tools so no real funds are involved.

Paste an address and get a plain report in about a second. Each line says what was checked and what it means.

Planned
Report
  • Mint authority revokedpass
  • Freeze authority revokedpass
  • Top 10 hold 39.0%pass
  • Liquidity lockedcheck
  • Metadata immutablepass
66Score
1Things to check
1.1sTook
Compare

Starter pack of 30 scenarios: prompt injection through token metadata, address swap in an invoice, inflated or repeated.

Line several tokens up side by side and sort by what matters to you.

Planned

Compare

TokenScoreTop 10Liquidity
LIN6565.2%locked
TOMI4419.0%unlocked
DOMI7840.5%locked
MORQ7657.3%locked
LORA8028.0%locked
Share

Per-scenario pass or fail with a replayable transcript showing exactly what the agent decided and why.

Every report has a link and an image you can post, so others see the same facts you saw.

Planned

Share card

KARA ยท 66/100Sparring report

The scenarios are specific to agents that move money on Solana and pay over x402. Generic LLM red-team tools test what a model says, while Sparring tests what the agent does with a wallet in hand.

How it works

3 steps from start to first result.

Paste an addressAny SPL token mint.
Read the reportEach check explained.
Share itLink or image.
Under the hood

Open by default.

Every screen reads from the same model. Live parts read public onchain data; nothing ever asks your wallet to sign.

Read the docs
sparring.json
{
  "product": "Sparring",
  "network": "solana",
  "features": [
    {
      "name": "Connect an agent by HTTP endpoint or MCP server,",
      "status": "planned"
    },
    {
      "name": "Starter pack of 30 scenarios: prompt injection t",
      "status": "planned"
    },
    {
      "name": "Per-scenario pass or fail with a replayable tran",
      "status": "planned"
    },
    {
      "name": "Shareable report page and badge showing which sc",
      "status": "planned"
    }
  ]
}

What works today.

Live features work today. Planned ones are next on the roadmap, in this order.

Connect an agent by HTTP endpoint or MCP server, with a mock wallet and mock RPC tools so no real funds are involvedPlanned
Starter pack of 30 scenarios: prompt injection through token metadata, address swap in an invoice, inflated or repeated.Planned
Per-scenario pass or fail with a replayable transcript showing exactly what the agent decided and whyPlanned
Shareable report page and badge showing which scenarios the agent passed and the date of the runPlanned
GitHub Action that reruns the pack on every release and fails the build on a regressionPlanned

Questions.

Short answers. The docs have the rest.

What is Sparring?

Run your onchain agent against a pack of traps (fake invoices, inflated 402 prices, poisoned token names) before it holds real funds. The scenarios are specific to agents that move money on Solana and pay over x402. Generic LLM red-team tools test what a model says, while Sparring tests what the agent does with a wallet in hand.

What works today?

Sparring is in development. Every feature is on the status list with what ships next, and the app opens in a sample workspace so you can see how it will work.

Do I need a wallet?

No. You can explore without one. Connecting a Solana wallet only shares your public address so Sparring can read your balances; it never asks you to sign a transaction.

Who is it for?

Teams shipping trading, payment or procurement agents, solo builders about to fund an agent for the first time, and funds that want evidence before they.

Connect an agent by HTTP endpoint or MCP server, with a mock wallet and mock RPC tools so no real funds are involved.

Run your onchain agent against a pack of traps (fake invoices, inflated 402 prices, poisoned token names) before it holds real funds.

Launch appRead the docs