io.github.aiguardrail由 GitHub 帳號 aiguardrail 發布v1.0.5更新於 Oct 24, 2026
Deterministic pre-trade risk checks for AI agents on Robinhood Chain. Returns allow/warn/deny.
一個遠端交易前風險檢查點，為 Robinhood Chain 上的擬議交易評分，並回傳 allow、warn 或 deny 及稽核雜湊。
Web executable，透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。
把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式，或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南，先向你展示發布者、端點和登入方式，經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。
閱讀 https://sourceweft.com/mcp/io.github.aiguardrail%2Fbackend/install.md，並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前，先告訴我它的發布者、端點和登入方式，等我確認。不要把 API 金鑰寫進命令或我會提交的檔案；問我金鑰應從哪裡取得。把它新增到你客戶端的 mcpServers 設定中。
{
"mcpServers": {
"collar-guardrail-pre-trade-risk-layer": {
"type": "http",
"url": "https://api.collarguardrail.com/mcp-http/mcp"
}
}
}Deterministic pre-trade risk layer for AI agents on Robinhood Chain.
Collar evaluates proposed trades before execution and returns
allow / warn / deny with reasons, a 0–100 risk score, and a
tamper-evident SHA-256 audit hash.
Collar sits between an autonomous trading agent and the chain as a deterministic pre-trade checkpoint. Every proposed trade is scored against live oracle prices, market-hours rules, tier-based limits, slippage bounds, and contract-safety signals before it is allowed to proceed.
The result is a single, auditable verdict — never a probabilistic guess.
https://api.collarguardrail.com/mcp-http/mcphttps://api.collarguardrail.com/.well-known/mcp-server-card.jsonhttps://api.collarguardrail.com/.well-known/agent-card.jsonhttps://api.collarguardrail.com/.well-known/x402.jsonhttps://api.collarguardrail.com/methodology.htmlhttps://api.collarguardrail.com/statushttps://api.collarguardrail.com/latency.htmlhttps://api.collarguardrail.com/llms.txthttps://collarguardrail.comMCP Transport: streamable-http | Protocol: 2025-06-18 | Auth: none (fixed Tier 1)
Add to your MCP client config (Claude Desktop, Cursor, VS Code):
來源：README.md，提交 8985c38
Evaluate Trade (Free)
Pre-trade risk check for a proposed trade. Call this BEFORE executing any trade and treat a "deny" decision as a hard stop. Evaluated at a fixed Tier 1 ceiling ($5,000 notional). For higher limits, use `evaluate_trade_paid` (Tier 2, $25,000) which settles an x402 payment on-chain.
side必填asset必填amount必填wallet必填request_idcontract_address必填max_slippage_bpsEvaluate Trade (x402 Paid, Tier 2)
Pre-trade risk check gated by an x402 payment. Evaluated as COLR Tier 2: $25,000 notional ceiling, 10/25 per-minute rate limits. No COLR holdings are required — the on-chain payment is the credential. FLOW ---- 1. Call WITHOUT payment_proof → the tool returns a `PaymentRequiredOutput` describing the on-chain payment (network, asset, amount, payTo, resource). 2. Settle the payment on-chain using an x402-capable client. The facilitator returns a proof string. 3. Call again with `payment_proof=<proof>` → the tool forwards the proof to the guardrail's x402-paid endpoint and returns a full `EvaluateTradeOutput` with `tier=2`.
side必填asset必填amount必填wallet必填request_idpayment_proofcontract_address必填max_slippage_bpsCheck Token Safety
Honeypot / contract safety check for any ERC-20 token. Returns a severity classification (safe / warn / danger), the specific risk factors detected, and the result of a simulated sell. When severity=danger, the next `evaluate_trade` for the same contract is automatically denied regardless of any other checks. This is a heuristic bytecode + simulation check, not a formal audit. Use it before trading unknown tokens, memecoins, or any contract that is absent from the official Robinhood Chain registry. It is safe to call repeatedly; no transaction is ever sent.
contract_address必填Simulate Balance
Simulate a wallet's ERC-20 balance after a hypothetical trade using eth_call state override. Read-only: no transaction is ever sent.
delta必填holder必填decimalstoken_address必填Get Supported Assets
List every asset in the guardrail's official Robinhood Chain registry, with its canonical contract address.
無需輸入參數。
Verify Audit Trail
Verify the tamper-evident audit trail for a wallet's past guardrail decisions.
limitwallet必填已滿足適用檢查項的 100%
該評分衡量此收錄條目的完整程度，依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價，也不表示伺服器安全或目前可用：可連通性是帶日期的檢查結果，評分也不包含安裝量或使用者評價。
{ "mcpServers": { "collar-guardrail": { "url": "https://api.collarguardrail.com/mcp-http/mcp" } }}Verify:
bashcurl -X POST https://api.collarguardrail.com/mcp-http/mcp \ -H "Content-Type: application/json" \ -H "Accept: application/json, text/event-stream" \ -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'Available ToolsTool Purposeevaluate_trade Pre-trade risk check at fixed Tier 1 ($5,000 ceiling). Returns allow / warn / deny with reasons and audit hash. Honeypot findings (severity=danger) force a DENY.evaluate_trade_paid Same as evaluate_trade, gated by an x402 payment. Evaluated as Tier 2 ($25,000 ceiling). Call without payment_proof to receive the challenge; settle on-chain; retry with the proof.check_token_safety Honeypot / contract safety check for any ERC-20. When severity=danger, evaluate_trade auto-DENYs.simulate_balance Read-only balance simulation via eth_call state overrideget_supported_assets Official Robinhood Chain asset registryverify_audit_trail Verify the hash-chain integrity of past decisionsCall get_supported_assets first to resolve a symbol to its canonicalcontract address. A mismatched address is treated as a fake-token attemptand denied.
The free evaluate_trade is capped at Tier 1 ($5,000). If you need Tier 2($25,000) without holding COLR, use evaluate_trade_paid — the on-chainx402 payment is the credential.
CapabilitiesPricing
Chainlink oracle prices for 34+ Stock Tokens on Robinhood Chain
Uniswap V4 fallback when the oracle is stale, paused, or unconfigured
No fabricated fallback — if neither oracle nor V4 yields a price, verdict is deny
10-second V4 price cache
US Market Hours
Full US holiday calendar with observed-on-Friday/Monday shift
Early close on Thanksgiving Friday and Christmas Eve
Policy Enforcement
Tier system (REST API): 1,000 / 2,500 / 5,000 COLR → $5K / $25K / $100K limits
Per-tier rate limiting
Cluster-based rate limit and daily-loss guard (keyed on wallet fingerprint)
Slippage / MEV guard (hard 500 bps floor)
Idempotency via request_id (5-minute cache)
Global kill switch, contract mismatch detection, per-wallet blocklist
Contract Safety
Honeypot detection: bytecode scan, owner check, sell simulation via eth_call
Auto-DENY: when severity=danger, the verdict is deny regardless ofother checks — a hard stop.
GoPlus intelligence for unregistered tokens (buy/sell tax, mintable,proxy, ownership reclaim).
Excessive tax (buy or sell > 10%) triggers an automatic deny viaGoPlus when allow_unregistered=true is passed to the request.
Auth & Infrastructure
JWT authentication via EIP-191 wallet signature (REST API)
Auth rate limiting per IP (20 req/min)
WalletConnect v2 + EIP-6963 wallet discovery
200+ token registry
Hash-chained audit trail (audit_hash + audit_seq)
Observability
Public live status page — uptime, dependency health, known issues
Public latency dashboard — p50/p95/p99 per endpoint and per MCP tool
Public methodology — how the 0–100 score is computed
Public audit-chain verification
Payments (x402)Endpoint: POST /api/x402/analyze
Price: $0.10 USDG per call
Network: Robinhood Chain (eip155:4663)
Discovery: /.well-known/x402.json
Facilitator: Ultravioleta DAO
Listing: x402 Bazaar
No signup, no API key. Registered in the x402 Bazaar for automatic agent discovery.
Response Formatjson{ "decision": "allow", "reasons": ["price source: oracle", "within tier 1 ceiling"], "tier": 1, "max_trade_usd": 5000, "calculated_notional_usd": 1452.30, "price_usd": 145.23, "price_source": "oracle", "risk_score": 12, "audit_hash": "sha256:...", "audit_seq": 12345, "daily_pnl_usd": 0.0}decision: allow / warn / deny — deny is a hard stop
reasons prefixed with ADVISORY: are non-blocking
price_source: oracle / uniswap_v4 / fallback_default / unavailable
error: if present, no verdict was produced — treat as a hard stop
Discovery Files/.well-known/mcp-server-card.json — MCP tool schemas (canonical, flat)
/.well-known/mcp/server-card.json — alias (nested)
/.well-known/agent-card.json — A2A skills
/.well-known/x402.json — x402 payment resources
/.well-known/methodology.md — risk score methodology (agent-readable)
/.well-known/known-issues.json — public known-issues list
/methodology.html — methodology, human-readable
/mcp-card.html — rendered server card
/agent-card.html — rendered agent card
/latency.html — live latency dashboard
/status — live status page
/llms.txt — full API contract for LLM agents
/skill.md — concise skill index
/robots.txt — AI crawler allow-list
/sitemap.xml — site map
Registry ListingsOfficial MCP Registry: io.github.aiguardrail/backend
Smithery: search "collar"
x402 Bazaar: /api/x402/analyze
DocumentationLive Terminal (UI): https://collarguardrail.com
Agent Integration Guide: https://collarguardrail.com/agent-docs.html
MCP Server Card: https://api.collarguardrail.com/.well-known/mcp-server-card.json
Methodology: https://api.collarguardrail.com/methodology.html
Status: https://api.collarguardrail.com/status
Latency: https://api.collarguardrail.com/latency.html
ContactX (Twitter): @collarguardrail
DisclosureCollar is the risk layer for autonomous finance on Robinhood Chain.Deterministic, auditable, and independent.
COLR is our own utility token, issued and operated by the Collar team.It is not affiliated with, endorsed by, or issued by Robinhood Markets,Inc. or any Robinhood entity. Robinhood Chain is a public blockchain webuild on — using it does not imply any relationship with Robinhood.