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.
-
Secure SSH access for AI agents via MCP. Execute commands across your server fleet with policy enforcement, network controls, and comprehensive audit logging.
Apache-2.0 ★ 30 -
A local, deterministic MCP evaluator for agent spend-policy eligibility. Returns ELIGIBLE, DENY, or STEP_UP with a stable reason code. Holds no funds, keys, or payment credentials, and makes no network requests.
npm MIT -
A deterministic policy firewall for AI agent tool calls. YAML rules, no LLM calls, no risk scores. CLI + MCP server (advisory) + @enforce() decorator (unbypassable). MIT, 46 tests, zero network dependency.
pypi MIT ★ 1 -
Analyse your Palo Alto Networks NetSec estate using CLI and MCP. Find configuration references and simulate Traffic to find out how data flows through your network and what policy rules actually match
MIT -
Governed, self-improving multi-agent swarm for 5G network operations on AWS: Bedrock agents, MCP tools behind AgentCore Gateway, Cedar policies, human approval, RL + federated learning on a digital twin. Finalist, AWS x FYUZ Breaking Barriers 2026 (team project).
-
Coragentic: an operating network for agents skills, private context, swarm coordination, jobs, policy, and proof on Robinhood Chain.
MIT -
@declaw/mcp-servergithub.com/declaw-ai/mcp-serverupdated this year
MCP server for Declaw — secure AI sandboxes with network policies, PII scanning, injection defense, and audit logging
npm Apache-2.0 -
Governed Postgres access for AI agents (MCP): RBAC policy, Vault secrets, signed single-use approvals, network allowlist + verify-full TLS, tamper-evident hash-chained audit.
MIT -
wardergithub.com/santosh-singhh/warderupdated this year
A policy gate for MCP tool calls. Blocks destructive AI agent actions before they run — as the subprocess your AI client launches, where network gateways can't see.
cargo MIT -
Travel assistant AI agent with a web UI — long-term memory (CUSTOM + SEMANTIC strategies), MCP Gateway + Tavily web search, per-agent Policy Group, network diagrams. Runs on GreenNode AgentBase.
MIT -
MCP-Mitigation---filesystem_executorgithub.com/BoldFaceType/MCP-Mitigation---filesystem_executorupdated this year
This is a compact MCP-compatible command gateway where MCP provides discovery and client compatibility, while a CLI/API execution substrate owns safe local-network execution. This gives up rich per-tool MCP schemas in exchange for low context cost, easier human debugging, and safer policy enforcement.
-
Eclipse Social Core MCP — the shared core of the Eclipse social adapter family: one canonical, hash-bound content package, per-adapter variants with classified diffs, hash-bound owner approvals, a capability-mode registry, and the engagement policy module. Governance only: no platform action, no network calls.
pypi MIT -
A supervised Hermes agent that plays the original 1991 DikuMUD through a custom Python MCP server driving TinTin++ over a PTY. Five tools, one command per turn enforced by a state machine, four least-privilege containers on per-flow networks, and a fixed-policy OpenRouter relay. The agent learns only from direct experience with the game world.
LGPL-2.1 -
Declaw -- Secure AI Sandboxesno update data
Secure runtime for AI agents. Run AI agents and their tools in Firecracker microVMs with network policies, PII scanning, prompt injection defense, and audit logging. 7 tools: create_sandbox, run_command, read_file, write_file, list_files, kill_sandbox, list_sandboxes. Security presets (none / standard / strict) let you control guardrails per sandbox: restrict outbound traffic to specific domains, redact PII from command output, block prompt injections, and get full audit trails.
smithery -
What's Allowedno update data
What your coding agent can do without asking you. Reads every settings file that contributes permission rules — managed policy, project, local, user — and reports the merged picture: which file each rule came from, which defaultMode wins, what proceeds unattended, and which rules the client accepts and then ignores (path rules on Write/Glob, unanchored allow globs, /path rules in user settings that anchor at the config directory, allow rules a deny rule reaches first). Four read-only tools: whats_allowed (full summary), permission_sources, rule_findings, unattended_surface. Read-only by construction — no child processes, no network, no writes; values of env entries are never read, only their names. MIT licensed. Free and MIT, with no telemetry and nothing held back. Built by Shift The Culture, who also publish paid kits for the failure modes this server surfaces: https://shifttheculture.media/agent-tools
smithery -
PrivacyScrubber PII Masking MCPno update data
Zero-Trust Data Sanitization (ZTDS) MCP Server for Claude Desktop, Cursor, Windsurf, and other AI clients. Automatically intercepts and redacts PII, credentials, and sensitive corporate data locally on your machine before they are sent to external LLMs. ### 🔒 100% Local & Airplane Mode Verified All sanitization runs inside local RAM. No data ever leaves your computer, ensuring compliance with GDPR, HIPAA, SOC 2, and internal GRC policies. Zero network calls, zero server logs. ### 🛠️ Supported Formats & File Types Locally parse and sanitize: - **Documents:** `.docx` (Microsoft Word) parsed locally with Zero-Server extraction. - **Code & Configs:** `.js`, `.py`, `.go`, `.rs`, `.json`, `.yaml`, `.ini`, `.env`, and source code files. - **Data & Logs:** `.txt`, `.log`, `.csv`, `.tsv`, `.md`, and plain text outputs. ### 🏷️ Redacted Entity Types - **Personal Data:** Names (`NAME`), Emails (`EMAIL`), Phone Numbers (`PHONE`). - **Credentials:** API Keys, passwords, tokens, SSH keys, database URIs. - **Identifiers (`ID`):** SSNs, Tax IDs, credit card numbers, and custom regex patterns. ### 🎛️ 22+ Specialized Industry Detection Profiles Specify detection rules tailored to your workload (requires PRO key): - **Engineering & IT:** `Dev`, `Tech`, `Security`, `Agents` - **Corporate & Operations:** `General`, `Bizops`, `Sales`, `Support`, `Marketing`, `Creative` - **Financial & Professional:** `Finance`, `WealthMgmt`, `Insurance`, `Accounting` - **Legal & Compliance:** `Legal`, `Compliance`, `CCPA`, `HR` - **Healthcare & Research:** `Medical`, `Pharma`, `Academic` - **Personal Use:** `Personal` ### 🔧 Available Tools - `sanitize_text` — Scrub PII/Secrets locally in text context. - `reveal_text` — Reversible token mapping (Original -> Token -> Original). - `sanitize_file` — Local sanitization of raw plain text and DOCX files.
smithery -
Browser-based GUI for Windows Server DNS — policies, zones, records, DNSSEC, DNS over HTTPS, and blocklists. Next.js frontend + PowerShell bridge, 8 scenario wizards, and an MCP server for AI agents. Works offline as a PowerShell command generator.
★ 8 -
Self-hosted UniFi Security Dashboard — firewall policy visualisation, IDS/IPS gap detection, policy drift, and MCP server for Claude Code integration
MIT ★ 3 -
Kubernetes security audit agent - declarative kagent agent with MCP tools and privacy-preserving pseudonymization via LLM proxy
Apache-2.0 ★ 1 -
AI-native chaos engineering orchestrator for Kubernetes — autonomous planning loop, DPv2-compatible engines (NetworkPolicy + Envoy fault injection), and MCP tools for arena/SUT/baseline.
Apache-2.0 ★ 2 -
Airline-booking-demo-strandsagent-agentcore-ver2.0-withsolutionmanagergithub.com/Anya2089/Airline-booking-demo-strandsagent-agentcore-ver2.0-withsolutionmanagerupdated this year
Airline customer service agent on AWS Bedrock AgentCore with dual-memory: conversation history + Solutions Memory for successful cases. Users click "Like" to save resolutions for future reference. Includes Gateway policy engine, MCP tools, and Cognito auth. Handles refunds, delay compensation, and loyalty rebooking.
★ 5 -
routeros-inspector-mcpgithub.com/kevinb361/routeros-inspector-mcpupdated this year
Unofficial read-only MCP server for MikroTik RouterOS inspection and policy-backed audits
MIT -
SatRailsgithub.com/darrylyeo/SatRailsupdated this year
Self-custodial, policy-governed BTC ↔ stablecoin routing and automations for agents and busy humans.
-
Nebulagithub.com/Caerlower/Nebulaupdated this year
Treasury infrastructure for AI agents on Stellar — custody, policy, liquid/yield split, and payment rails.
MIT ★ 7 -
An MCP server that gives AI agents access to public internet interconnection data: who peers with whom, where, and under what policy.
pypi MIT -
Clean and explore sensitive data by talking to it. A WebMCP workspace where the browser refuses to let the page make a network request.
MIT -
MCP server: Cisco Foundation-Sec-8B detects and blocks SDN attacks with simulated, auto-rolled-back policy (IOS-XE RESTCONF/NETCONF)
Apache-2.0 -
mcp-fortigate-publicgithub.com/tempestcha/mcp-fortigate-publicupdated this year
MCP server for querying FortiGate firewalls via FortiOS REST API. Read-only tools for firewall policies, routing, VPN, UTM events, and local syslog — for Pi, Claude Code, and other MCP clients.
MIT -
pci-dss-compliance-mcpgithub.com/crawde/pci-dss-compliance-mcpupdated this year
MCP server for PCI DSS v4.0 compliance — browse requirements, assess compliance readiness, generate policies, evidence checklists, and gap analysis for organizations handling payment card data
npm MIT -
MCP server for Cisco Catalyst SD-WAN Manager (vManage). Read-only by default: fabric inventory, device health, control connections, BFD/OMP state, alarms, tunnel path quality, templates and policies as MCP tools for Claude and other AI assistants.
pypi MIT ★ 1 -
Automate Cisco Secure Firewall (FMC-managed FTD) with Python, Ansible, and Terraform, plus three MCP servers that let an AI agent search policy, trace object usage, and propose changes behind a preview-and-confirm gate. Read-only by default, TLS on by default, CSV-driven, lab-tested.
MIT