Token-safety for AI agents: rug/honeypot verdict + score across PulseChain, Monad, Base & BSC
MCP (Model Context Protocol) server that gives any AI agent on-chain token-safety tools for PulseChain, Monad, Base, and BSC — including PulseChain, which the major token-safety APIs (GoPlus, Honeypot.is) don't support at all (verified June 2026). If your agent touches PulseChain, this is the safety check.
It is a thin stdio client over the hosted API at onchain.wick.pics
(engine wick-safe/0.3). Every check is deterministic on-chain analysis — no LLM in the
verdict path, so results are fast, cheap, and reproducible: contract-risk heuristics,
liquidity depth, honeypot transfer-simulation, LP-burn, ownership/renounce + privilege
detection, upgradeable-proxy detection, and a wallet approval (drainer) scanner. The free tier
needs no key. Informational, not financial advice.
| tool | what it answers |
|---|---|
check_token_safety({ chain, address }) | Is this token a scam? Verdict (SAFE … LIKELY_RUG), 0–100 score, evidence checks. |
fresh_rug_radar() | What just launched, and is it safe? Last 20 freshly-created pools, safety-scored at creation. |
exit_safety({ chain, token, sizeUsd, maxSlippage? }) | Could I sell $X of this at acceptable slippage? Size-aware price impact + safety verdict. |
check_ownership({ chain, address }) | Is ownership renounced / upgradeable, and what can an active owner still do (mint, blacklist, pause, tax)? |
safe_to_interact({ chain, address }) | One call → SAFE_TO_INTERACT / CAUTION / DO_NOT_INTERACT, bundling safety + ownership, with reasons. |
wallet_approvals({ chain, owner }) | A wallet's active ERC-20 approvals, flagging unlimited grants — the drainer vector. |
wallet_poison_check({ chain, owner }) | Address-poisoning scan: dust/look-alike transfers in a wallet's history that mimic a real counterparty to trick a paste-and-send of the wrong address. |
chain: pulsechain | monad | base | bsc · token/address: 0x… contract address.
Agents should call check_token_safety / exit_safety before interacting with, buying, or
accepting an unknown token.
{
"mcpServers": {
"onchain-safety": {
"command": "npx",
"args": ["-y", "github:aitools420/onchain-safety-mcp"],
"env": { "ONCHAIN_API_KEY": "wsk_…" }
}
}
}
ONCHAIN_API_KEY is optional — omit it for the free rate-limited tier. A wsk_ key
unlocks the paid deep tier (get one at onchain.wick.pics).
The same engine is available with no signup and no key via the x402 pay-per-call protocol — agents pay USDC on Base per request. Machine-readable index of every paid route (token-safety, fresh-rug radar, exit-safety):
https://onchain.wick.pics/.well-known/x402.json
Human docs: onchain.wick.pics/agents
| var | default | purpose |
|---|---|---|
ONCHAIN_API_BASE | https://onchain.wick.pics | API host |
ONCHAIN_API_KEY | (none — free tier) | wsk_ key for the paid deep tier |
MIT
Source-derived launch command. Check the maintainer’s required arguments and credentials before running:
npx -y @wickpics/onchain-safety-mcpMerge this template into ~/Library/Application Support/Claude/claude_desktop_config.json. Keep existing servers. Add any arguments, credentials, and permissions required by the maintainer; this template has not been install-tested.
{
"mcpServers": {
"io-github-aitools420-onchain-safety-mcp": {
"command": "npx",
"args": [
"-y",
"@wickpics/onchain-safety-mcp"
]
}
}
}Restart Claude Desktop completely for changes to take effect. Confirm the server appears connected in the client’s tool list, then try a read-only example from its documentation.
Claude Desktop setup reference@wickpics/onchain-safety-mcpnpmio.github.aitools420/onchain-safety-mcp works with any MCP-compatible client. Copy the config snippet from the Configuration section above and add it to the file shown for your client, then restart the application.
~/Library/Application Support/Claude/claude_desktop_config.jsonRestart Claude Desktop completely for changes to take effect.~/.cursor/mcp.jsonRestart Cursor for changes to take effect..vscode/mcp.jsonReload VS Code window for changes to take effect.~/.codeium/windsurf/mcp_config.jsonRestart Windsurf for changes to take effect..mcp.jsonSave at the project root, then start Claude Code in that project and review the MCP server approval prompt. Keep real credentials out of shared files.