Loading ToolBench
Mantle Network MCP server for reading state, estimating gas, managing RWA, and checking agent reputation.
Mantle-MCP has 19 tools with significant definition gaps. Most tools lack complete input schemas visible in source (tools/advanced_execution.py imports are unresolved). Descriptions vary widely: some are detailed (get_native_balance, estimate_l2_gas), others are minimal or missing (get_mantle_network_info, get_token_address have empty input schemas with no descriptions). Parameter descriptions are present for Phase 1 tools but sparse for Phase 2 wrappers. No output schemas documented. Error handling is absent, tools return raw exceptions or generic dicts. Security concern: no validation of wallet addresses, token symbols, or amounts. Composition is poor: tools like get_rebalance_bundle, get_arbitrage_bundle, get_autocompound_bundle are imported from unresolved modules, making their actual schemas unverifiable. STDIO transport caps protocol readiness at 50.
Estimates the gas units and total cost required to execute a transaction on Mantle L2.
Fetches the current mETH exchange rate from the Ethereum L1 Mantle LSP Contract.
Generates an unsigned transaction bundle to approve and supply an asset to Aave V3 on Mantle.
Generates a cross-DEX arbitrage execution bundle between two Mantle DEXes.
Generates a bundle to claim pending rewards and zap them back into the principal LP.
Gets a V3 swap quote (exact input) from specific DEXes like 'merchant_moe' or 'agni_finance'.
Retrieves the ERC-20 token allowance granted by a wallet to a specific spender.
Missing input schemas for 7 tools (get_mantle_network_info, get_token_address, fetch_meth_yield, scan_aave_yields, moe_zap_bundle, get_rebalance_bundle, get_arbitrage_bundle, get_autocompound_bundle). LLMs cannot infer parameter types or constraints.
No output schemas documented for any tool. Agents cannot plan downstream calls or extract required fields (e.g., does scan_aave_yields return APY, pool_address, token_symbol?).
No error handling or recovery guidance. Tools return raw exceptions (e.g., 'Exception' in fetch_meth_yield) or generic dicts with 'error' key. LLMs cannot determine if errors are retryable or user-fixable.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 58 | <=2025-11-25 | v2 |
Retrieves the ERC-20 token balance for a specific wallet address.
Retrieves the native MNT balance of an address on the Mantle Network.
Generates a transaction bundle to rebalance an uneven portfolio by swapping on a target protocol.
Gets a swap quote (expected output) for a given trade pair on a DEX using Uniswap V2 Router logic.
Generates a single atomic transaction to zap native MNT into the Merchant Moe WMNT/USDT LP on Mantle Mainnet.
Returns an unsigned payload to register and mint an ERC-8004 Identity NFT on the Mantle Network.
Returns an unsigned EVM payload to wrap USDC into USDY on the Mantle Network.
Queries an agent's on-chain reputation score on the Mantle Network.
Scans the Aave V3 on-chain contract for real-time asset supply APYs and returns the top opportunities.
Checks the owner address of an ERC-8004 Agent Identity on the Mantle Network.
No input validation. Tools accept wallet_address, token_symbol, amount_wei without checking format, range, or validity. LLMs can pass invalid EVM addresses or negative amounts.
Unresolved imports in tools/advanced_execution.py (generate_rebalance_bundle, generate_arbitrage_bundle, generate_autocompound_bundle). Actual schemas and implementations are not visible in source, making verification impossible.