Loading ToolBench
18 pay-per-call x402 services for AI agents — on-chain anchoring & attestation, wallet/address security screening, Web3 data (tx & calldata decode, ENS, token prices), x402 spend accounting, content analysis, and verifiable randomness. No API keys or accounts; settle per request in USDC on Base or Solana.
Critical gaps across all dimensions. Tool names lack action verbs (anchor, screen, attest, decode_tx, decode_calldata, resolve_name, token_price, intel_wallet, parse_datetime, chat, roast, oracle, tldr, aura, grade, roll). Descriptions are present but generic and lack WHEN/WHY context. Input schemas visible but parameters lack descriptions in most tools. No output schemas documented. Error handling absent. Composition issues: 'chat', 'roast', 'aura', 'grade', 'oracle', 'tldr', 'roll' are vague utility tools that don't follow verb_noun pattern. 'anchor' and 'attest' are domain-specific but lack clarity on prerequisites and failure modes.
Accept a hash (or arbitrary JSON to be hashed), write the resulting 32-byte digest to Base mainnet (calldata) and Solana mainnet (Memo program) in parallel, return both tx hashes.
Verify an ephemeral eip191 signature and dual-chain anchor the decision receipt on Base + Solana mainnet.
Analyze the aura or vibe of a wallet based on its transaction patterns and holdings.
Chat endpoint for conversational AI interactions.
Decode EVM calldata hex string into function name and parameters.
Decode a blockchain transaction by hash. Returns transaction details including block number, from/to addresses, value, and function calls.
Tool names lack action verbs. 'anchor', 'screen', 'attest', 'chat', 'roast', 'oracle', 'tldr', 'aura', 'grade', 'roll' do not follow verb_noun pattern. LLMs cannot infer intent from names alone.
No output schemas documented. LLMs cannot plan downstream calls or extract required fields. Responses are unstructured.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 45 | 2026-07-28+ | v2 |
Grade or score a wallet address on various metrics.
Get intelligence and risk metrics for a wallet address including transaction history, balance, and activity patterns.
Query an oracle for real-time data or predictions.
Parse a datetime string in various formats and return normalized ISO 8601 format.
Resolve an ENS name to an Ethereum address or vice versa.
Generate a humorous roast of a wallet address based on its on-chain activity.
Generate verifiable randomness via on-chain randomness oracle.
Sanctions + AML screen for a Base or Ethereum wallet address. Returns sanctions_match, sanctioned_lists, and a risk_level tier.
Generate a TLDR (too long; didn't read) summary of text content.
Get the current price of a token by symbol or contract address.
Parameter descriptions missing or minimal. 'hash', 'wallet', 'chain', 'tx_hash', 'calldata', 'name', 'symbol', 'message', 'text', 'seed', 'query' lack format/constraint guidance. LLMs cannot validate inputs.
No error handling guidance. Tools do not document retryable vs fatal errors, recovery steps, or actionable error messages. Agents cannot self-correct.
Vague utility tools ('chat', 'oracle', 'tldr', 'roast', 'aura', 'grade', 'roll') lack clear composition. No indication of when to use each vs alternatives. Agents will struggle to select the right tool.