AI agents and MCP explained
Plain answers about MCP, agent tools and the real cost of AI work, with sources and limits.
Remote vs local MCP servers: hosting, authorization and which one to pick
Choose between local stdio and remote HTTP MCP servers, understand OAuth authorization, and verify an editor connection before granting access.
Guide · ENLocal models with MCP: run an agent with Ollama, and know its limits
Connect Ollama to MCP tools through a local agent client. Learn the setup, hardware constraints, privacy boundaries and where Tvijo fits.
Guide · ENMCP in VS Code: add a server, trust it, and audit its actions
Add an MCP server in VS Code, choose the right JSON format, understand trust prompts, and check tool evidence. Includes the boundaries of Tvijo’s editor clients.
Guide · ENHow much does an AI coding agent cost per task? Tokens, limits and free starts
Understand token-based costs, subscription limits and free previews. Calculate an illustrative task price and check Tvijo’s free-start and prepaid-credit conditions.
Guide · ENMCP server security checklist: 12 checks before connecting real data
Verify MCP servers, narrow permissions, test prompt injection boundaries and require review before connecting an agent to real data.
Guide · ENMCP vs API vs function calling: which does your AI integration need?
Use a direct API for fixed requests, function calling for model-selected actions, and MCP for tools shared across compatible clients. Learn the trade-offs and safe setup.
Guide · ENWhat is MCP? Servers, clients, tools and resources in plain words
Model Context Protocol connects AI applications to tools and data. Learn what servers and clients do, how transports differ, and how to check an editor connection.