Skip to content
InvokeIndex

Every tool an agent can call

50,869 public listings across 7 directories, each with its source and history. 15,879 more from bulk publishers wait for review.

Scan an MCP serverUpdated 3 Oct, 16:07 UTC. How we collect

Smithery (11,302)

Page 26 of 471

Latest changes

Typed diffs between observations
  1. ChainRay Official MCP Registry wallet.credit_score: description changed (+"Scope: wallet-level financial behavior and activity score based on transaction history, balances, token diversity, stablecoins, and DeFi use. Use this for 'how active or creditworthy is this wallet?'. It is not Agent identity/reputation (agent.reputation), a registered card (agent.card.lookup), or token/contract risk.")
  2. ChainRay Official MCP Registry util.verify_proof: description changed (+"Use after receiving a ChainRay response with _proof.dataHash to verify its integrity. It verifies a supplied hash and does not independently validate the underlying blockchain data.")
  3. ChainRay Official MCP Registry util.stream_info: description changed (+"Use to discover the five real-time SSE streams and their connection requirements. The returned guidance is not a stream itself; connect directly over HTTP with the documented Accept and payment headers.")
  4. ChainRay Official MCP Registry util.intent_router: description changed (+"Use when the agent has a natural-language data question and needs ChainRay to select a focused route. For deterministic workflows, call the specific tool directly; paid HTTP requests are separate from this MCP capacity policy.")
  5. ChainRay Official MCP Registry util.context_brief: description changed (+"Use at the start of a swap, bridge, invest, monitor, or general workflow to collect the relevant signals in one structured brief. Use focused tools when the agent already knows the exact data it needs.")
  6. ChainRay Official MCP Registry token.supply: description changed (+"Use for contract-level supply metadata such as total supply, decimals, symbol, and name. It does not assess price, token safety, or holder concentration; use token.risk, token.price, or token.holders instead.")
  7. ChainRay Official MCP Registry token.risk: description changed (+"Scope: asset-level token due diligence for a token contract or mint, combining market, liquidity, verification, and control signals. Use this when the question is 'is this token safe or liquid?'. It is not raw bytecode inspection (use security.contract.scan), a wallet's financial behavior (use wallet.credit_score), or a multi-signal address dashboard (use analyze.risk_dashboard).")
  8. ChainRay Official MCP Registry token.holders: description changed (+"Use for token-holder concentration and recent net accumulation. Do not use it for a wallet's portfolio or token price; use wallet.report for a wallet address or token.price for a token price.")
  9. ChainRay Official MCP Registry security.sandwich: description changed (+"Use to inspect historical sandwich patterns around an address. It is not a pre-trade protection decision; use security.mev.shield before a swap or liquidity action.")
  10. ChainRay Official MCP Registry security.mev.shield: description changed (+"Scope: composite pre-trade decision for a proposed swap, bridge, or liquidity action, focused on MEV exposure and slippage mitigation. Use this before acting. It is not a historical sandwich scan (security.sandwich), a token due-diligence report (token.risk), or a transaction executor.")
  11. ChainRay Official MCP Registry security.contract.scan: description changed (+"Scope: bytecode-level inspection of one EVM contract for proxy, ownership, delegatecall, and selfdestruct signals. Use this for 'what does the deployed code expose?'. It is not a token market/liquidity assessment (token.risk), a wallet score, or a full audit, and it does not judge a trade by itself.")
  12. ChainRay Official MCP Registry monitor.watch.diff: description changed (+"Use only after monitor.watch.address has created a watchId in the same MCP session. It returns changes since the last check and advances the snapshot; create a new watch for a new address.")

Coverage

Brands resolved
22,185
On two or more hosts
5,088
Remote MCP servers
24,263
Tools listed by probe
8,395 servers

Not listed yet?

Scan any public MCP server against both directories’ review rules.