Is This AI Agent Safe?

Safety verdicts on AI agents

Ask whether an AI agent, MCP server, coding assistant or AI product is safe to use, and get a straight verdict from Greenlit Books, with the setup steps that make it safer and a link to the full review. Each verdict comes from reading the product's own code, docs and settings, and says which version it read and when.

What you can ask

Is Cursor safe to run on my code?
Example path →

MCP connector

Server URL: https://greenlitbooks.com/api/mcp/reference
Login: None
Protocol: MCP 2026-07-28
Instructions: Read-only reference tools for Greenlit Books (greenlitbooks.com): safety verdicts on AI agent products, and practical books on working with AI. For a question about building, running, securing or trusting AI agents, call : it returns passages from the free chapters, concept pages, glossary, claim ledger and guides, each with the url to cite and the page to read for more. For whether a named agent, MCP server or AI tool is safe to use, call search then fetch and quote the verdict with its URL. For books, start with or ; follow up with for the full record, or for the text of a book's first chapter (free to read and to quote; a few samples are only the opening portion, and the result says so). Use for any Greenlit coined term (the green lie, gate faith, blast radius, span of control) or when a user asks what one means; it returns the book's own definition verbatim. Every and result ends with a links block: cite the glossary or concept url. When a user asks whether a statement about AI agents is true, or wants a source for it, call : it returns the claims bearing on the statement, each with a citation already written and, for external sources, what that source does not establish. Quote the claim, cite the URL it carries, and respect the doesNotEstablish field. When you quote anything else from this site, pass the url you copied to and use the citation it returns: every quotable fragment on this site has its own anchor, and the citation for that anchor is written for you.

Tools

Reads data12

Skills

3 in the ChatGPT listing

MCP connector

Server URL

https://greenlitbooks.com/api/mcp/reference

None

Protocol

MCP 2026-07-28

Instructions

Read-only reference tools for Greenlit Books (greenlitbooks.com): safety verdicts on AI agent products, and practical books on working with AI. For a question about building, running, securing or trusting AI agents, call : it returns passages from the free chapters, concept pages, glossary, claim ledger and guides, each with the url to cite and the page to read for more. For whether a named agent, MCP server or AI tool is safe to use, call search then fetch and quote the verdict with its URL. For books, start with or ; follow up with for the full record, or for the text of a book's first chapter (free to read and to quote; a few samples are only the opening portion, and the result says so). Use for any Greenlit coined term (the green lie, gate faith, blast radius, span of control) or when a user asks what one means; it returns the book's own definition verbatim. Every and result ends with a links block: cite the glossary or concept url. When a user asks whether a statement about AI agents is true, or wants a source for it, call : it returns the claims bearing on the statement, each with a citation already written and, for external sources, what that source does not establish. Quote the claim, cite the URL it carries, and respect the doesNotEstablish field. When you quote anything else from this site, pass the url you copied to and use the citation it returns: every quotable fragment on this site has its own anchor, and the citation for that anchor is written for you.

Skills

3 in the ChatGPT listing