Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
作者提供的说明(英文原文)
AgentGuard — 20-tool AI safety MCP: policy preflight, risk scoring, audit logging, rate limits.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
适用于网页版、桌面版和手机版 Claude:
https://feedoracle.io/guard-oracle/mcp/在终端中运行这条命令
claude mcp add --transport http agentguard https://feedoracle.io/guard-oracle/mcp/粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"agentguard": {
"url": "https://feedoracle.io/guard-oracle/mcp/"
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"agentguard": {
"type": "http",
"url": "https://feedoracle.io/guard-oracle/mcp/"
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
ChatGPT 只能连接互联网上的服务器:
https://feedoracle.io/guard-oracle/mcp/菜单名称可能因套餐和版本不同而略有差异。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
我们还没有读取这个服务器的工具列表;我们会轮流检查各个端点。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月6日
注册名称已通过域名 tooloracle.io 验证(注册中心会核实域名所有权)
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
尚未检查;我们会轮流检查各个端点
尚未检查;我们会轮流检查各个端点
尚未检查;我们会在检查端点时读取工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。
GitHub 仓库最后更新于–,最新版本(1.0.0)于6个月前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。
同一分类中有:Spec Driven Development, Remote Desktop Commander, HotCodePush
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Manage over-the-air (OTA) updates for React Native, Expo, Capacitor and Cordova apps.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
Security, Policy & Audit Layer for AI Agent Tool Execution — 20 tools | Part of FeedOracle & ToolOracle
AgentGuard is the governance and security layer for AI agent workflows. Before any tool executes, AgentGuard evaluates policies, scores risk, detects secrets and injection attempts, logs to a tamper-evident audit trail, and explains every decision. Built for regulated environments, autonomous payments, and enterprise AI agent deployments.
# FeedOracle (compliance-focused)
npx -y mcp-remote https://feedoracle.io/guard/mcp/
npx -y mcp-remote https://tooloracle.io/guard/mcp/
{
"mcpServers": { "agentguard": { "command": "npx", "args": ["-y", "mcp-remote", "https://feedoracle.io/guard/mcp/"] } } }
Agent wants to call: payment_execute({amount: 5000})
│ ▼ policy_preflight() ← Check before execution │ risk_score: 95 ← Critical matched: pol-001, pol-002 ← Payment + High-risk policies decision: require_approval ← Human gate triggered │ approval_required() ← Register pending approval │ [Human approves] │ tool executes │ audit_log_write() ← Record outcome with signature │ decision_explain() ← Exportable compliance evidence
| Tool | Description |
|---|---|
policy_preflight | Pre-flight check before any tool call. Evaluates 7 policies, computes risk score, detects threats, auto-logs. Returns allowed/denied/require_approval/flagged. |
tool_risk_score | 0-100 risk score for tool + input. Detects secrets (API keys, passwords), prompt injection, high-value amounts. eth_gas→5, payment_execute→95+. |
approval_required | Check if tool needs human approval. Optionally registers a pending approval request with tracking URL. |
audit_log_write | Write tool execution to persistent, cryptographically-signed audit log (SQLite WAL). Call after execution to record outcome. |
audit_log_query | Query audit trail. Filter by agent, tool, decision, time range. Paginated. Returns signed entries for tamper verification. |
decision_explain | Human-readable explanation of any allow/deny decision. Pass request_id for stored entry or tool_name + tool_args for fresh analysis. |
rate_limit_check | Check agent rate limits: 200/min, 5000/hr, 50000/day. Returns per-window usage with percentage. |
| Tool | Description |
|---|---|
payment_policy_check | Validate payment against policy: amount limits (>100k warns, >1M blocks), recipient denylist, supported currencies/networks, AML thresholds (>10k fiat flagged), MiCA flags. |
spend_limit_check | Check per-call/hour/day spend limits by trust level. Default: 10k/call, 50k/hr, 200k/day. Trusted: 100k/call, 500k/hr, 2M/day. |
secret_exposure_check | Deep scan for 19 secret patterns: OpenAI/GitHub/AWS/Slack keys, Bearer/Basic auth, ETH private keys, Bitcoin WIF, credit cards, SSNs, emails. Returns severity + remediation. |
payload_safety_check | 18-pattern safety scan: prompt injection, jailbreak/DAN, role hijacking, SQL (UNION/DROP/OR 1=1), XSS, Python/JS/Shell injection, path traversal, null bytes, oversized payloads. |
replay_guard_check | Detect replay attacks via SHA256 fingerprint (agent+tool+args). Configurable window (default 5 min). Returns duplicate count + first/last seen. |
| Tool | Description |
|---|---|
cross_tool_anomaly_check | Detect anomalous patterns: risky combos (wallet-recon→transfer, AML→payment), high frequency, repeated denials (policy probing), broad reconnaissance, elevated avg risk score. |
scope_check | Role-based scope control. Roles: admin, compliance_officer, trader, auditor, developer, readonly. Returns has_scope, missing scope, granting roles. Logs denials. |
session_validate | Full session lifecycle: create (TTL + call budget), validate (increment counter), invalidate, info. Sessions carry role, scopes, tenant, expiry. |
tenant_policy_check | Multi-tenant governance. Built-in tenants: default, fintech_eu (MiCA/DORA/AMLD6), defi_protocol, enterprise_read. Per-tenant blocklists, risk limits, spend caps. |
threat_intel_check | Entity threat intelligence. Auto-detects ETH addresses, IPs, domains. Checks sanctions (Tornado Cash, mixers), disposable services, behavioral analysis from audit log. |
| Tenant | Max Risk | Spend/Day | Frameworks |
|---|---|---|---|
default | 70 | 100,000 | — |
fintech_eu | 60 | 500,000 | MiCA, DORA, AMLD6 |
defi_protocol | 80 | 10,000,000 | MiCA |
enterprise_read | 30 | 0 | — |
| Role | Scopes |
|---|---|
admin | All scopes |
compliance_officer | audit:read, compliance:read, blockchain:read, security:scan |
trader | blockchain:read, payment:check, payment:execute, audit:read |
auditor | audit:read, audit:write, compliance:read, monitor:read |
developer | blockchain:read, security:scan, audit:read, monitor:read |
readonly | blockchain:read, audit:read |
| Policy | Condition | Action |
|---|---|---|
| pol-001 | Payment/transfer tools | require_approval |
| pol-002 | Risk score ≥ 80 | require_approval |
| pol-003 | Secret/key in payload | deny |
| pol-004 | Rate limit exceeded | flag |
| pol-005 | Risk score ≤ 20 | allow freely |
| pol-006 | Prompt injection detected | deny |
| pol-007 | Same tool > 50 calls/60s | flag |
| Score | Level | Action |
|---|---|---|
| 0-14 | Minimal | Proceed freely |
| 15-39 | Low | Proceed, log for audit |
| 40-69 | Medium | Flag and proceed with caution |
| 70-89 | High | Require human approval |
| 90-100 | Critical | Block execution |
approval_required gate before executionPersistent, stable, no daemon required. WAL-mode supports 1000+ writes/second. Shared between feedoracle.io and tooloracle.io — one source of truth.
https://feedoracle.io/guard/mcp/https://tooloracle.io/guard/mcp/*Built by FeedOracle — Evidence by Design*