Sponge Wallet

WarnAudited by ClawScan on May 10, 2026.

Overview

This appears to be a real crypto wallet integration, but it can sign/send transactions and contains a risky shell-based login helper, so it should be reviewed carefully before use.

Install only if you trust Sponge/Paysponge and are comfortable giving this skill transaction-signing authority. Use testnet or low-balance wallets first, keep recipient allowlists and spending limits tight, confirm every transfer/swap/payment manually, and avoid setting SPONGE_API_URL unless you fully trust the endpoint.

Findings (5)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

If the skill or agent is invoked incorrectly, it may have enough authority to sign transactions or move crypto assets.

Why it was flagged

The OAuth login requests delegated wallet read/write and transaction signing authority, which can control funds once an API key is stored or provided.

Skill content
scope: "wallet:read wallet:write transaction:sign transaction:write"
Recommendation

Use only with a trusted Sponge account, prefer testnet or low-balance wallets, enforce allowlists/spending caps, and require explicit user confirmation before any transfer, swap, withdrawal, or signed payment.

What this means

A mistaken prompt, misunderstood recipient, or over-autonomous invocation could send funds, swap tokens, or incur paid API charges.

Why it was flagged

The skill exposes irreversible financial mutation tools as direct tool calls, while the instructions do not specify a mandatory final confirmation or transaction preview before execution.

Skill content
`evm_transfer` | Transfer ETH/USDC on Ethereum/Base ... `solana_transfer` | Transfer SOL/USDC on Solana ... `solana_swap` | Swap tokens via Jupiter
Recommendation

Before installing or using, ensure the agent must restate and confirm chain, recipient, amount, token, slippage, fees, and payment cost before every transaction or paid API call.

What this means

A compromised provider response or untrusted SPONGE_API_URL could cause local shell commands to run on the user's machine during authentication.

Why it was flagged

The login flow interpolates a server-returned device code into a shell command. JSON.stringify is not shell escaping, so a malicious or redirected auth endpoint could potentially trigger command execution during login.

Skill content
execSync(`echo -n ${JSON.stringify(device.userCode)} | pbcopy`);
Recommendation

Replace shell clipboard commands with a safe clipboard API or spawn commands with fixed arguments and stdin; validate the device code format before copying.

What this means

The provider can see authenticated wallet actions and paid API prompts, URLs, or search parameters submitted through the skill.

Why it was flagged

Wallet commands and bearer credentials are sent to a remote MCP endpoint. This is expected for the integration, but the endpoint receives sensitive wallet operations and paid API task data.

Skill content
const MCP_ENDPOINT = `${API_URL}/mcp`; ... "Authorization": `Bearer ${apiKey}`
Recommendation

Only use trusted endpoints, avoid submitting sensitive prompts or private document URLs unless intended, and do not set SPONGE_API_URL to an untrusted server.

What this means

It may be harder to verify that the installed skill came from the expected wallet provider.

Why it was flagged

The registry metadata does not provide strong provenance for a high-impact wallet skill, even though the README references a GitHub repository.

Skill content
Source: unknown; Homepage: none; No install spec — this is an instruction-only skill.
Recommendation

Verify the repository and script contents from a trusted Sponge/Paysponge source before granting wallet access.