MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
Holds risky AI agent tool calls for a person, ties each approval to one exact request, and logs every decision. 210 checks, synthetic data.
MIT -
Pre-sign safety check for AI agent wallets and MCP: decode a transaction, EIP-712 Permit / Permit2 / EIP-3009 signature, EIP-7702 delegation or x402 payment and flag drainer patterns (unlimited or EOA approvals, setApprovalForAll, fake USDC domains, x402
npm MIT -
bonanza-mcpgithub.com/c6zks4gssn-droid/bonanza-labs-websiteupdated this year
MCP server for AI agent payments — spending firewall, budget checks, and approval flows via the Model Context Protocol
pypi Apache-2.0 -
consentgate-mcpgithub.com/jessepetersondev/consentgate-mcpupdated this year
MCP server: gate AI-agent actions behind a consent policy + human approval (Telegram). Tools: check_action, request_approval. Fail-closed.
npm MIT -
Phygital Key for MCP: AI apps ask through MCP, a person approves the exact action with a passkey, any service checks the signed yes offline. Built on Phygital-Key/phygital-key.
npm Apache-2.0 -
truss-ap-automation-mcpgithub.com/claw-factory/truss-ap-automation-mcpupdated this year
TRUSS AP Automation MCP — AI-powered accounts payable automation for agents. Invoice extraction, GL coding, duplicate detection, approval routing, compliance checks, and payment scheduling.
npm MIT -
@spendctl/mcp-serveractive
MCP server exposing spendctl to agents: pay, await_approval, withdraw_approval, check_budget, list_receipts. Holds no keys.
npm SEE LICENSE IN LICENSE -
Typed inventory tools with capability checks and order approval
MIT -
ParityDiff MCP: public API contract evidence, local call-site checks and explicitly approved migration drafts
npm Apache-2.0 -
Kubernetes MCP server with human approval: AI proposes, you approve each change with a passkey from your phone. Alerts, recovery check, audit log.
oci Apache-2.0 ★ 2 -
kitlistgithub.com/kingbedjed/kitlistupdated this year
A human-approval gate between an AI agent and your Amazon cart — the agent proposes a shopping list, the human approves, Amazon checks out. MCP tool + llms.txt.
-
io.github.gosswerner1-max/lazyrelay-mcp-servergithub.com/gosswerner1-max/lazyrelay-integrationsno update data
Publish posts to every platform, check platform rules, manage drafts and approvals, read analytics.
npm -
Open-source privacy and approval gateway between AI assistants and your business systems (MCP): human approval, local PII checks and audit for Gmail, Drive, Slack, Salesforce, Jira and more.
pypi Apache-2.0 ★ 2 -
Limetry is open-source agent action governance: check tool intents against policy before irreversible runs. Returns allow, deny, or approval_required with signed receipts and privacy-safe audit. TypeScript SDK, CLI, and MCP for IDE like Cursor, VS Code, GitHub Copilot, Claude. Optional CI, SQL, and Shopify adapters. MIT — self-host your own node.
npm MIT ★ 1 -
The model proposes, deterministic code checks every order against a versioned YAML mandate, and a human approves; approval never bypasses the guard. Autonomous paper trading agent on Alpaca with least privilege MCP tools
MIT -
mtrkr-mcp-servergithub.com/n1n4du/mtrkr-mcp-serverupdated this year
MCP server for MTRKR wallet intelligence on MegaETH — token scanning, approval checking, tx decoding, .mega names
npm ★ 4 -
MCP tool-call gate for CloakPipe: contract checks, approvals, side-effect tracking and evidence on every tools/call
cargo MIT -
MCP server that checks who released a contract before an agent signs a payment, approval or account upgrade.
npm MIT -
Safe, auditable Kubernetes actions for AI agents: plan, policy-check, approve, apply.
Apache-2.0 -
A read-only MCP server for bounded readiness checks of operator-approved service dependencies.
MIT -
Verify before connect: check the MCP tools your client receives against your approvals and LEGION's signed log (Python, MIT)
MIT -
Typed MCP tools with allowlisted actions, approval checks, audit events, optional n8n webhook dispatch, and in-memory protocol tests.
MIT -
Local control panel and MCP server for building, checking and publishing small-business websites with human approval gates.
NOASSERTION -
Follow the line: read Ethereum transactions one stop at a time, check an address before you pay it, and check your wallet for risky approvals. Web, terminal and MCP for AI agents.
MIT -
Local MCP server for writing consistent serial fiction with continuity checks and approval workflows, then adapting chapters into webtoon scenes: one image per scene with dialogue in the work language, pre-generation checks and visual review.
npm Apache-2.0 -
github-deploy-mcpgithub.com/kartikanand73/github-deploy-mcpupdated this year
MCP server that lets an AI agent trigger, check, and analyze GitHub Actions deploys — in plain language, without bypassing your existing approval gates.
MIT -
Catch the wrong invoice before delivery: deal checks, AI agents over MCP, and PayPal dispute replies built only from proof the seller approves.
MIT -
MCP server that checks whether a detection rule will actually work in your environment, and enables it only with human approval.
Apache-2.0 -
AI agent for routine SAP Basis user admin: standard BAPIs via an MCP server, with human approval, SoD check and audit log (POC)
MIT -
Remote approvals, policy checks, and execution evidence for unattended AI agents.
npm MIT ★ 318 -
Keeps your Windows PC's RAM and disk in check: monitoring, safe cleanup with approval, a desktop pet, and an MCP server.
pypi MIT -
Execute local or remote host commands through MCP with policy checks, optional TUI approval, live output, and persisted results.
Apache-2.0 ★ 1 -
Local-first, human-approves-everything grocery replenishment as a ResonantOS add-on: runway analysis, fail-closed dietary checks, and a next-cart draft that structurally cannot check out. No accounts, no cloud, no telemetry — nothing leaves your machine.
MIT -
PerfoAds MCP: a free, self-hosted MCP server for Google Ads. 162 checks, changes only with your approval. Not a Google product.
npm Apache-2.0 ★ 3 -
MCP server: ERC-20 honeypot and rug-pull check on Base and Ethereum before your agent swaps or approves, plus swap route data.
MIT -
Agent Bodyguard — 5 onchain safety tools (copycat detection, sandwich alarms, exit-liquidity checks, deployer tracking, approval audits) as an MCP server and x402 API. Built for the PayBox/MoonPay hackathon.
-
营销文案事实审校与视频合成 Agent:Qwen 经 MCP 检索资料逐句核查,缺证据即停止;5 种检索策略 Hit@5 10%→43.3%,4B/14B 对照留出集 75%→100%;148 测试、OTEL/Prometheus 可观测、人工确认绑定快照后渲染 FFmpeg 视频。 | Marketing fact-checking agent: Qwen retrieves evidence over MCP, refuses unsupported rewrites; 5 retrieval strategies, 4B/14B comparison, 148 tests, OTEL metrics, version-bound approval, FFmpeg rendering.
MIT ★ 1 -
Method, synthetic sheet and checks for letting an AI agent place native takeoff measurements in Bluebeam through MCP. Every quantity stays a proposal until a person approves it.
MIT ★ 2 -
AGORA+: a self-hosted household assistant. Memory with receipts, approval before action, a team of agents that checks plans against the family's facts, and one MCP server for Alexa+, Claude and ChatGPT.
MIT -
A voice-controlled care necklace for seniors living alone. ESP32-S3 firmware, Python MCP tools and a Next.js caregiver dashboard for check-ins and activity approvals. Top 5 Finalist, Dell InnovateDash 2026.
NOASSERTION -
Research data desk for a university finance lab: find a dataset, check it can really be downloaded, approve, collect and register it. React interface, REST API and MCP. Interface and reference runtime; demo included.
-
Transaction security for EVM chains. ShieldBot checks tokens, approvals and signatures before a wallet signs, says what it checked, and reports Unknown instead of Safe when evidence is missing. Browser extension, Telegram bot, REST API, MCP server and SDKs. Covers Ethereum, BNB Chain, opBNB, Base, Arbitrum, Polygon, Optimism and Robinhood Chain.
MIT ★ 2 -
AtlasBurn MCP server: ask Claude, Cursor or any MCP client where your AI spend goes, check guardrails and runaway-agent signals, and set AtlasBurn up with your approval.
NOASSERTION -
Spending guard for AI agents that pay x402 paywalls in USDC: budgets, payee allowlists, price-spike checks and human approval before anything is signed. MCP server. Arc + NEAR testnet.
npm MIT -
MCP server for the x402-research pay-per-call API (crypto signals/briefs, Base address risk, x402 invoice/settlement/approval checks). Testnet by default; your own wallet pays, capped per call and per session.
npm MIT -
Agentic backend service (FastAPI + LangGraph + MCP) with an orchestrator routing to specialized agents, a 3-node agentic loop, security-guarded MCP tools (SQL/SSRF/path-traversal checks), and human-in-the-loop approval for sensitive operations.
MIT -
Kizuki (review-pulse) — turns Groww app reviews into a weekly product pulse and a fact-checked support answer, published via MCP only after a human approves.
-
PalPluss MCP server — let AI coding tools check balances, manage payment channels and wallets, collect with M-Pesa STK Push and send B2C payouts, with human approval on anything that moves money.
npm MIT -
kontrol-freek-mcpgithub.com/berkbayri/kontrol-freek-mcpupdated this year
AI assumption firewall for MCP agents — intercepts risky decisions, checks past decisions, detects contradictions, and gates human approval via Telegram or Slack.
pypi ★ 1 -
ephor-gateupdated this year
Ephor gate — the identity & policy layer for AI agents. Standard MCP gateway (stdio + HTTP): every agent action risk-scored, policy-checked, human-approved when it matters, and recorded in a tamper-evident audit chain. https://ephor.dev
npm SEE LICENSE AT https://ephor.dev -
Zero-trust gateway for AI agent tool calls: policy-checked, approval-gated, and recorded in post-quantum signed receipts anyone can verify offline.
MIT -
QA for any running website: an MCP server for AI hosts (Claude Code, Antigravity, Gemini CLI, Cursor) and a standalone command, with test-account logins, real-browser actions, check packs, evidence, bug reports and approved Nashmi cards
npm MIT -
Safely connect Google Ads to Grok, Grok Bot and Cursor, so your agent can check campaigns, audit search terms, research keywords and make the changes you approve. Built for PPC specialists, agencies and in-house teams.
MIT -
Human-approved AI triage for a 401(k) service desk: PII tokenization, call-type classification, plan and payroll record checks, routing, and a code-enforced close guard. MCP servers and a an AI agent(Claude).
-
Safely connect Microsoft Ads (Bing Ads) to Grok, Grok Bot and Cursor, so your agent can check campaigns, find wasted spend, research keywords and make the changes you approve. Built for PPC specialists, agencies and in-house teams.
MIT -
Buy now, wait, or split it into 4? A cash-flow check before you pay: a deterministic engine, Claude to read the question, and a PayPal sandbox checkout you approve yourself.
MIT -
Safely connect LinkedIn Ads to Grok, Grok Bot and Cursor, so your agent can check campaigns, audiences and lead forms, flag what needs attention and make the changes you approve. Built for B2B marketers, agencies and in-house teams.
MIT -
Mishmar is a risk management layer for trading bots and AI agents: every order is checked against your limits before it reaches the exchange, large orders wait for your approval, and no key can trade alone.
npm Apache-2.0 -
basescopegithub.com/chasdaddy/basescopeupdated this year
The read-only safety layer for onchain AI agents — an MCP server for safe, read-only onchain checks (token/rug safety, risky approvals, verified source, balances, ENS, gas, prices) on Base + EVM. No private keys, no required API keys.
MIT -
@varynttools/work-mcpactive
MCP server for Varynt Work (https://work.varynt.in/), project and client management for agencies — lets an AI assistant read and create tasks, log time, pull reports and check approvals using your own API token.
npm MIT