# MCP Server Security > Search a sourced MCP risk register, work through a security checklist and compare 14 tools that find and control MCP servers on developer machines. Every page has a markdown twin: add index.md to its address, or send Accept: text/markdown. Full text of every page: https://mcpservercybersecurity.com/llms-full.txt ## home - [MCP Server Security: Risk Register and Tool Comparison](https://mcpservercybersecurity.com/): Search a sourced MCP risk register, track a security checklist in your browser and see 14 MCP discovery and control tools scored on public evidence. ## ranking - [MCP Server Security Tools Ranked: 14 Scored on Evidence](https://mcpservercybersecurity.com/ranking/): Fourteen tools that discover and control MCP servers on developer machines, scored 0 to 100 on seven weighted criteria from public vendor evidence. ## method - [How MCP Security Tools Are Scored: Method and Weights](https://mcpservercybersecurity.com/method/): One research plan for every vendor, 733 facts in 14 vendor files, each quote checked against its page, with blind scoring and weights set first. ## vendors-index - [MCP Security Vendors: 14 Tool Profiles](https://mcpservercybersecurity.com/vendors/): Profiles of 14 tools that discover and control MCP servers on developer machines, each with scores, reasons and the sourced facts behind them. ## vendor - [Backslash Security MCP Security: Endpoint MCP Proxy](https://mcpservercybersecurity.com/vendors/backslash-security/): How Backslash Security discovers MCP servers on endpoints, blocks risky tool calls and runs a free server vetting hub, with scores. - [Bay MCP Security: Endpoint Scan and Enforcer](https://mcpservercybersecurity.com/vendors/bay/): How Bay scans MCP servers before approval and enforces tool-call decisions on the device, with scores, sources and a stated weakness. - [Cisco MCP Security: MCP Scanner and AI Defense](https://mcpservercybersecurity.com/vendors/cisco/): How Cisco's open-source MCP Scanner and AI Defense hooks scan MCP servers and check tool calls, with scores and sourced facts. - [CrowdStrike MCP Security: Falcon Guardian](https://mcpservercybersecurity.com/vendors/crowdstrike/): How CrowdStrike Falcon Guardian discovers MCP servers and malicious servers on endpoints and links MCP use to execution, with scores. - [Harmonic Security MCP Security: Local MCP Gateway](https://mcpservercybersecurity.com/vendors/harmonic-security/): How Harmonic Security's endpoint agent and local gateway inventory MCP servers and block tool calls, with scores and sourced facts. - [Knostic MCP Security: Kirin and AgentMesh](https://mcpservercybersecurity.com/vendors/knostic/): How Knostic's Kirin endpoint client and AgentMesh scanner rate MCP servers and block unapproved connections, with scores and facts. - [Koi MCP Security: Cortex Agentic Endpoint Security](https://mcpservercybersecurity.com/vendors/koi/): How Koi and Palo Alto Networks Cortex discover MCP servers, IDE extensions and packages on Windows and macOS, with scores. - [Lasso Security MCP Security: Discovery and Scanning](https://mcpservercybersecurity.com/vendors/lasso-security/): How Lasso Security inventories MCP servers, scans tool descriptions and monitors tool calls with SIEM export, with scores and facts. - [Microsoft MCP Security: Defender for Endpoint](https://mcpservercybersecurity.com/vendors/microsoft/): How Microsoft Defender for Endpoint discovers local AI agents and their MCP servers on Windows and macOS, with scores and sourced facts. - [Noma Security MCP Security: Registry and Controls](https://mcpservercybersecurity.com/vendors/noma-security/): How Noma Security inventories MCP servers, runs an approval registry and restricts tools through hooks, with scores and sourced facts. - [Pillar Security MCP Security: Agentic Endpoint](https://mcpservercybersecurity.com/vendors/pillar-security/): How Pillar Security inventories MCP servers, scores them against a catalog and enforces per-server tool allow-lists, with scores and facts. - [SentinelOne MCP Security: Prompt Security](https://mcpservercybersecurity.com/vendors/sentinelone/): How SentinelOne's Prompt Security discovers and risk-scores MCP servers and logs agent actions, with scores and sourced facts. - [Snyk MCP Security: Agent Scan and Evo ADS](https://mcpservercybersecurity.com/vendors/snyk/): How Snyk Agent Scan and Evo ADS find, vet and block MCP servers on developer machines, with scores and the facts behind them. - [Zenity MCP Security: Live Tool Call Control](https://mcpservercybersecurity.com/vendors/zenity/): How Zenity's MCP Security product shows MCP requests and tool calls and can allow, modify or block them, with scores and sourced facts. ## bestfor-index - [Best MCP Security Tools by Situation](https://mcpservercybersecurity.com/best-for/): Five situations re-rank the same evidence: many developers, a fast inventory, an approved catalog, stopping tool calls and a regulated audit. ## bestfor - [Best MCP Security Tool for Many Developers](https://mcpservercybersecurity.com/best-for/many-developers/): Which MCP server tool suits hundreds of developers adding servers to IDEs. Discovery weighs 30 and deployment 20, with scores from public evidence. - [Best MCP Security Tool for a Fast MCP Inventory](https://mcpservercybersecurity.com/best-for/fast-inventory/): Which MCP server tool gives a security team an inventory across machines this month. Discovery weighs 40 and deployment 30 in this view. - [Best MCP Security Tool for an Approved Server Catalog](https://mcpservercybersecurity.com/best-for/approved-catalog/): Which MCP tool helps a platform team keep an approved list of vetted servers. Vetting weighs 40 in this view, from public evidence. - [Best MCP Security Tool to Stop Tool Calls](https://mcpservercybersecurity.com/best-for/stop-tool-calls/): Which MCP tool can allow, block or hold an MCP tool call on the machine. Runtime control weighs 40 in this view, from public evidence. - [Best MCP Security Tool for a Regulated Audit](https://mcpservercybersecurity.com/best-for/regulated-audit/): Which MCP tool shows which servers ran and what they did. Governance and audit weighs 30 in this view, scored from public evidence. ## asset - [MCP Risk Register: 42 sourced MCP server risks](https://mcpservercybersecurity.com/mcp-risk-register/): A sourced register of 42 MCP server risks in six dimensions, with impact, controls, severity and primary sources, read on 16 Nov 2026. - [MCP Security Checklist: 42 steps in six stages](https://mcpservercybersecurity.com/mcp-security-checklist/): A checklist of 42 concrete steps to discover, vet, authenticate, constrain, observe and respond to MCP servers, mapped to risks and sources. ## asset-method - [MCP Risk Register method and sources](https://mcpservercybersecurity.com/mcp-risk-register/methodology/): How the MCP Risk Register is built: what it holds, which sources it reads, the severity rule, how to cite it, the license and how to report corrections. - [MCP Security Checklist method and sources](https://mcpservercybersecurity.com/mcp-security-checklist/methodology/): How the MCP Security Checklist is built: its sources, the priority rule, how to cite it, the CC BY 4.0 license and how to report corrections. ## asset-dim - [MCP supply chain risks: register](https://mcpservercybersecurity.com/mcp-risk-register/supply-chain/): Seven sourced MCP supply chain risks with controls and severity: look-alike packages, launchers, dependencies, rug pulls, registries and unknown servers. - [MCP authorization risks: register](https://mcpservercybersecurity.com/mcp-risk-register/authorization/): Eight sourced MCP authorization risks with controls and severity: token audience, open HTTP endpoints, proxy consent, issuers, scopes and shared state. - [MCP transport and network risks: register](https://mcpservercybersecurity.com/mcp-risk-register/transport-and-network/): Seven sourced MCP transport and network risks with controls: DNS rebinding, open listeners, SSRF, redirects, TLS, request limits and egress. - [MCP tool descriptions and schemas risks: register](https://mcpservercybersecurity.com/mcp-risk-register/tool-descriptions-and-schemas/): Six sourced MCP risks in what the model reads: poisoned descriptions, shadowing, schemas, annotations, tool results and hostile schemas. - [MCP execution and filesystem risks: register](https://mcpservercybersecurity.com/mcp-risk-register/execution-and-filesystem/): Seven sourced MCP execution risks with controls: user privileges, startup commands, injection, path traversal, project config and auto-approval. - [MCP data exposure and logging risks: register](https://mcpservercybersecurity.com/mcp-risk-register/data-exposure-and-logging/): Seven sourced MCP data risks with controls: environment secrets, token files, exfiltration, elicitation, missing logs and secrets in logs. - [MCP security checklist: discover](https://mcpservercybersecurity.com/mcp-security-checklist/discover/): Seven checklist steps to find MCP clients and servers on machines, with controls and sources. - [MCP security checklist: vet](https://mcpservercybersecurity.com/mcp-security-checklist/vet/): Eight checklist steps to vet an MCP server before use, with controls, advisories and primary sources. - [MCP security checklist: authenticate](https://mcpservercybersecurity.com/mcp-security-checklist/authenticate/): Seven checklist steps to authenticate MCP callers and tokens: audience, scopes, issuers and storage, with sources. - [MCP security checklist: constrain](https://mcpservercybersecurity.com/mcp-security-checklist/constrain/): Eight checklist steps to constrain MCP servers: allow lists, sandboxes, loopback binding, egress and approvals, with sources. - [MCP security checklist: observe](https://mcpservercybersecurity.com/mcp-security-checklist/observe/): Six checklist steps to log and monitor MCP tool calls and definition changes, with controls and sources. - [MCP security checklist: respond](https://mcpservercybersecurity.com/mcp-security-checklist/respond/): Six checklist steps to respond to a bad or vulnerable MCP server: remove, rotate, preserve, scope, patch and learn. ## datasets-index - [MCP Security Datasets: Risk Register and Checklist](https://mcpservercybersecurity.com/datasets/): Download the MCP Risk Register and the MCP Security Checklist as CSV and JSON under CC BY 4.0, with a citation line for each edition. ## dataset - [MCP Risk Register dataset, CSV and JSON](https://mcpservercybersecurity.com/datasets/mcp-risk-register/): Download the MCP Risk Register from MCP Server Security as CSV or JSON under CC BY 4.0, with the cite line and the date read. - [MCP Security Checklist dataset, CSV and JSON](https://mcpservercybersecurity.com/datasets/mcp-security-checklist/): Download the MCP Security Checklist from MCP Server Security as CSV or JSON under CC BY 4.0, with the cite line and the date read. ## guides-index - [MCP Server Security Guides: Twelve Practical Guides](https://mcpservercybersecurity.com/guides/): Twelve sourced guides on securing MCP servers: best practices, vetting, review, catalogs, scanning, three languages, cloud, logging and response. ## guide - [Building an Approved MCP Server Catalog for Your Company](https://mcpservercybersecurity.com/guides/building-an-approved-mcp-server-catalog/): How to run a company MCP server catalog: entry fields, approvers, enforcement in Claude Code and Cursor, and re-review triggers. - [How to Review an MCP Server for Security: Code and Tools](https://mcpservercybersecurity.com/guides/how-to-review-an-mcp-server-security/): A security review routine for an MCP server: read the tool list and code, test authorization, inputs and transport, and write a report. - [How to Vet an MCP Server Before Use: Steps and Signals](https://mcpservercybersecurity.com/guides/how-to-vet-an-mcp-server-before-use/): A pre-use vetting routine for MCP servers: publisher, source, startup command, tool descriptions and scopes, with the signals that mean stop. - [MCP Audit Logging: What to Record and How Long to Keep It](https://mcpservercybersecurity.com/guides/mcp-audit-logging-what-to-record/): Which fields an MCP audit log needs, what to redact, where to send logs and why retention must be checked, sourced to OWASP and vendor docs. - [MCP Security Best Practices for Server Operators](https://mcpservercybersecurity.com/guides/mcp-security-best-practices/): Eleven MCP security best practices for developer machines and remote endpoints, sourced to the MCP specification and OWASP, with a checklist map. - [MCP Security Incident Response: Steps for a Bad Server](https://mcpservercybersecurity.com/guides/mcp-security-incident-response-steps/): What to do when an MCP server is found malicious or vulnerable: contain, find every machine, revoke tokens and confirm cleanup, with sources. - [MCP Security vs API Security: What Changes](https://mcpservercybersecurity.com/guides/mcp-security-vs-api-security/): How MCP security differs from API security: model-chosen calls, shared tool descriptions and local servers, and which API controls still apply. - [MCP Server Scanning and Testing: Tools and Limits](https://mcpservercybersecurity.com/guides/mcp-server-scanning-and-testing/): Which scanners check MCP servers, what they miss, where to run them and how to test a server you build, with open-source options named. - [MCP Server Security on AWS, Azure and Google Cloud](https://mcpservercybersecurity.com/guides/mcp-server-security-on-aws-azure-and-google-cloud/): What AWS AgentCore Runtime, Azure API Management and Google Cloud Run provide for hosted MCP servers, and which settings stay with you. - [Securing a Python MCP Server: Host Checks and Deployment](https://mcpservercybersecurity.com/guides/securing-a-python-mcp-server/): Fix the Invalid Host header 421 error in the Python MCP SDK without switching off protection, and set transport security for deployment. - [Securing a Spring Boot MCP Server: Spring AI Settings](https://mcpservercybersecurity.com/guides/securing-a-spring-boot-mcp-server/): How to add OAuth 2.0 to a Spring AI MCP server, protect tools with method security and check which settings expose the endpoint. - [Securing a TypeScript MCP Server: Input, Binding and Tokens](https://mcpservercybersecurity.com/guides/securing-a-typescript-mcp-server/): Validate tool input with schemas, bind an Express MCP server to localhost with host checks and verify bearer tokens by audience in the TypeScript SDK. ## glossary-index - [MCP Security Glossary: 28 Terms Defined](https://mcpservercybersecurity.com/glossary/): Plain definitions of 28 MCP security terms, from tool descriptions and server pinning to DNS rebinding, each with a sourced example. ## term - [Command injection: MCP security glossary](https://mcpservercybersecurity.com/glossary/command-injection/): Command injection is when untrusted text is run as an operating system command. In MCP it appears in tools and in clients that open server-supplied URLs. - [DNS rebinding: MCP security glossary](https://mcpservercybersecurity.com/glossary/dns-rebinding/): DNS rebinding lets a web page talk to a service on your own machine by changing what a domain name points to. Local MCP servers on HTTP are exposed. - [Dynamic Client Registration: MCP security glossary](https://mcpservercybersecurity.com/glossary/dynamic-client-registration/): Dynamic Client Registration lets an OAuth client register itself with an authorization server on the fly. MCP deprecated it in 2026-07-28. - [Localhost binding: MCP security glossary](https://mcpservercybersecurity.com/glossary/localhost-binding/): Localhost binding means a local server listens only on 127.0.0.1, so other machines cannot connect. The MCP spec recommends it for local HTTP servers. - [MCP authorization: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-authorization/): MCP authorization is the optional OAuth-based scheme for protecting remote MCP servers. The spec sets token audience and discovery rules. - [MCP config file: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-config-file/): An MCP config file lists the servers a client launches or connects to, with the commands and environment variables. Editing it changes what code runs. - [MCP elicitation: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-elicitation/): MCP elicitation is how a server asks the user for input through the client. Form mode must never collect passwords or API keys. - [MCP Inspector: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-inspector/): The MCP Inspector is the reference tool for testing MCP servers, with web, command-line and terminal clients. It holds credentials and starts processes. - [MCP Registry: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-registry/): The MCP Registry is the official catalog of metadata for public MCP servers, with DNS-verified namespaces. It points to code and does not host it. - [MCP resource: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-resource/): An MCP resource is data a server exposes to a client by URI, such as a file or a schema. Policies on resources need their own checks. - [MCP roots: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-roots/): MCP roots are the directories a client tells a server it cares about. The spec calls them guidance, not access control. - [MCP sampling: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-sampling/): MCP sampling lets a server ask the client's model for a completion. The 2026-07-28 spec deprecates it, with removal no earlier than July 2027. - [MCP session: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-session/): An MCP session was connection state in older protocol revisions. The 2026-07-28 revision has no protocol-level sessions, so servers mint handles. - [MCP transport: MCP security glossary](https://mcpservercybersecurity.com/glossary/mcp-transport/): An MCP transport is the channel that carries MCP messages. The standard ones are stdio for local servers and Streamable HTTP for remote ones. - [npx launcher: MCP security glossary](https://mcpservercybersecurity.com/glossary/npx-launcher/): The npx launcher runs a package from npm on demand. Many local MCP servers start this way, so the package name in the command decides what executes. - [Package supply chain: MCP security glossary](https://mcpservercybersecurity.com/glossary/package-supply-chain/): The package supply chain is every npm, PyPI or container package an MCP server pulls in. Each one is code that runs with the server's permissions. - [Path traversal: MCP security glossary](https://mcpservercybersecurity.com/glossary/path-traversal/): Path traversal is when a file path lets a tool reach files outside the folder it should be limited to. MCP file tools are a frequent place for it. - [Protected resource metadata: MCP security glossary](https://mcpservercybersecurity.com/glossary/protected-resource-metadata/): Protected resource metadata (RFC 9728) is the document an MCP server publishes to say which authorization servers can issue tokens for it. - [Resource indicators: MCP security glossary](https://mcpservercybersecurity.com/glossary/resource-indicators/): Resource indicators (RFC 8707) let an OAuth client name the server a token is for. MCP clients must send one so tokens carry the right audience. - [Server manifest: MCP security glossary](https://mcpservercybersecurity.com/glossary/server-manifest/): A server manifest is the metadata file that says how to find and run an MCP server. In the MCP Registry it is called server.json. - [Server pinning: MCP security glossary](https://mcpservercybersecurity.com/glossary/server-pinning/): Server pinning means fixing an MCP server to a reviewed version, and often a hash of its tool definitions, so a later change cannot slip in. - [Server provenance: MCP security glossary](https://mcpservercybersecurity.com/glossary/server-provenance/): Server provenance is the evidence of who built an MCP server, where it came from and whether the code you run matches it. - [SSRF: MCP security glossary](https://mcpservercybersecurity.com/glossary/ssrf/): SSRF (server-side request forgery) tricks a program into fetching a URL of an attacker's choosing, often an internal address or a cloud metadata service. - [Tool annotations: MCP security glossary](https://mcpservercybersecurity.com/glossary/tool-annotations/): Tool annotations are hints a server attaches to a tool, such as read-only or destructive. Clients must treat them as untrusted by default. - [Tool description: MCP security glossary](https://mcpservercybersecurity.com/glossary/tool-description/): A tool description is the text an MCP server gives an agent to explain a tool. It is read by the model, so treat it as untrusted input. - [Tool schema: MCP security glossary](https://mcpservercybersecurity.com/glossary/tool-schema/): A tool schema is the JSON Schema that defines the arguments an MCP tool accepts. A strict schema narrows what a model can send. - [Tool shadowing: MCP security glossary](https://mcpservercybersecurity.com/glossary/tool-shadowing/): Tool shadowing is when one MCP server's tool text steers how an agent uses tools from other, trusted servers. Learn the risk and the control. - [Typosquatted server: MCP security glossary](https://mcpservercybersecurity.com/glossary/typosquatted-server/): A typosquatted MCP server is a package with a name one keystroke from a trusted one, published in the hope someone installs it by mistake. ## news-index - [MCP Security News: Vulnerabilities and Spec Changes](https://mcpservercybersecurity.com/news/): Twelve dated items on MCP server vulnerabilities, malicious packages, specification and registry changes, and release notes from MCP security tools. ## news - [Knostic: MCP tool asks agents for a GitHub token](https://mcpservercybersecurity.com/news/knostic-mcp-tool-description-asks-agent-for-github-token/): Knostic research (12 Nov 2026) on @agentdance/node-webrtc-mcp 1.0.13, whose tool description tells agents to fetch a GitHub token without asking. - [MCP draft spec adopts Client ID Metadata draft 03](https://mcpservercybersecurity.com/news/mcp-spec-draft-client-id-metadata-document-draft-03/): The MCP draft specification (9 Nov 2026) updates its Client ID Metadata Document reference to draft 03; implementations that support it must follow it. - [MCP Registry v1.8.8: key listings by name and version](https://mcpservercybersecurity.com/news/mcp-registry-v1-8-8-listings-keyed-by-name-and-version/): MCP Registry v1.8.8 (6 Nov 2026) notes that different server names may share a remote URL, so consumers should key listings by name and version. - [MCP Python SDK client fetched URLs from server schemas](https://mcpservercybersecurity.com/news/python-sdk-client-fetched-server-chosen-ref-urls/): GHSA-rwrf-2pqf-9j8j (27 Oct 2026): the MCP Python SDK client resolved server-chosen $ref URLs when validating tool output. Fixed in 1.30.0 and 2.2.0. - [Knostic flags a repackaged Amazon SP-API MCP server](https://mcpservercybersecurity.com/news/knostic-repackaged-amazon-sp-api-mcp-credential-redirect/): Knostic research (26 Oct 2026): npm package @bestcoolman/dev-mcp copies Amazon's SP-API MCP but sends credentials to a hardcoded IP. - [FastMCP advisories: authorization skipped after transforms](https://mcpservercybersecurity.com/news/fastmcp-authorization-advisories-transforms-and-issuer-binding/): Two FastMCP advisories on 23 Oct 2026: component authorization skipped after server transforms, and OAuth client wrappers without issuer binding. - [MCP TypeScript SDK server advisories on audience](https://mcpservercybersecurity.com/news/mcp-typescript-sdk-server-advisories-audience-memory-cross-client/): Four GitHub advisories on 20 Oct 2026 for MCP TypeScript SDK servers: token audience, shared transport message mix-up, memory use and experimental tasks. - [Cisco DefenseClaw 1.0.0 adds managed installs and inventory](https://mcpservercybersecurity.com/news/cisco-defenseclaw-1-0-0-managed-enterprise-release/): Cisco's open-source DefenseClaw reached 1.0.0 on 11 Oct 2026: managed installs, identity-based profiles, IDE and plugin inventory, ten agents. - [MCP TypeScript SDK OAuth client advisory (CVE-2026-104850)](https://mcpservercybersecurity.com/news/mcp-typescript-sdk-oauth-client-credentials-advisory/): GitHub advisory CVE-2026-104850: the TypeScript SDK's OAuth client could send stored credentials to a server-chosen authorization server. Fixed in 1.31.0. - [Snyk makes Evo ADS MCP Governance generally available](https://mcpservercybersecurity.com/news/snyk-evo-ads-mcp-governance-generally-available/): Snyk says Govern Agent Behavior is generally available, starting with MCP Governance: approved server lists, logging or blocking on the endpoint. - [mcp-atlassian: 25 advisories, one critical, fixed in 0.22.0](https://mcpservercybersecurity.com/news/mcp-atlassian-25-advisories-authentication-bypass-fixed-0-22-0/): GitHub published 25 advisories for the mcp-atlassian server on 22 Sep 2026, including a critical HTTP authentication bypass fixed in 0.22.0. - [MCP Python SDK 2.2.0 tightens redirects and sessions](https://mcpservercybersecurity.com/news/python-sdk-2-2-0-redirect-session-issuer-defaults/): Release notes for MCP Python SDK 2.2.0: same-origin redirects, idle session expiry, issuer checks and a token audience setting. ## faq - [MCP Server Security FAQ: Risks, Scanners and Tools](https://mcpservercybersecurity.com/faq/): Short sourced answers on whether MCP servers are a security risk, how to secure and check them, whether scanners exist and which tools are open source. ## about - [About MCP Server Security](https://mcpservercybersecurity.com/about/): What MCP Server Security publishes: a sourced risk register, a checklist, twelve guides and a scored comparison of tools that control MCP servers. ## editorial-policy - [Editorial Policy: How MCP Security Pages Are Written](https://mcpservercybersecurity.com/editorial-policy/): How facts are sourced, how vendor claims are labeled, how scores are set and when pages change on MCP Server Security. ## corrections - [Corrections and Tips: Report an Error or Source](https://mcpservercybersecurity.com/corrections/): How to report a factual error on MCP Server Security, how corrections are checked and logged, and how to send a tip about an MCP issue. ## contact - [Contact MCP Server Security](https://mcpservercybersecurity.com/contact/): Email the editor with questions, corrections, tips or data requests for MCP Server Security. One address, one reply. ## privacy - [Privacy Policy for MCP Server Security](https://mcpservercybersecurity.com/privacy/): What data MCP Server Security collects: CloudFront access logs kept 90 days, no cookies, no analytics, and checklist progress in local storage only. ## terms - [Terms of Use for MCP Server Security](https://mcpservercybersecurity.com/terms/): Terms for MCP Server Security: the CC BY 4.0 license for the datasets, the citation format, read dates on every fact and how to report an error. ## Data - [MCP Risk Register (CSV)](https://mcpservercybersecurity.com/datasets/mcp-risk-register.csv) - [MCP Risk Register (JSON)](https://mcpservercybersecurity.com/datasets/mcp-risk-register.json) - [MCP Security Checklist (CSV)](https://mcpservercybersecurity.com/datasets/mcp-security-checklist.csv) - [MCP Security Checklist (JSON)](https://mcpservercybersecurity.com/datasets/mcp-security-checklist.json) - [News feed](https://mcpservercybersecurity.com/feed.xml)