Claude Code / Reference
Claude Code and MCP
MCP allows Claude Code to connect to external tools and data sources.
What does MCP solve?
MCP lets Claude Code connect to external tools and data sources, such as documents, issues, database read-only queries, design systems, internal APIs, or services outside the browser.
It's good for pulling the context of "Claude is not in the code base" into the conversation, but it also expands the scope of permissions. Each MCP tool should have clear purpose, permissions, and logs.
Configuration scope
MCP servers can be configured per user, project, or locally. Project-level configuration is suitable for team sharing; user-level configuration is suitable for personal tools; local configuration is suitable for machine-related or sensitive services.
| range | suitable for |
|---|---|
| user level | Commonly used personal tools and accounts. |
| Project-level .mcp.json | Team sharing tools that accompany the repository into a web session. |
| local coverage | The connection is machine dependent or should not be submitted. |
Permissions and security
MCP tools may read or write to external systems. Only connect to the servers required by the task, and set approval or deny rules for write operations, delete operations, and production data access.
- Read-only queries take precedence over write tools.
- Production tools must be auditable.
- Tool results also go into context, subject to control over size and sensitive fields.
Complete usage points
Supplement the core concepts, operation sequences, permission boundaries and verification requirements that are easily compressed and missed in official documents, making it easier for English readers to learn completely by page.
MCP Supplement
MCP allows Claude Code to connect to external tools and data sources such as documentation, GitHub, Issues, database read-only queries, design systems, monitoring, and internal APIs. It solves the "context outside the code base" problem and also introduces new permission boundaries.
Configuration scope can be user-level, project-level, or local overrides. Project-level .mcp.json is suitable for team sharing and entering web sessions with the repository; user-level is suitable for personal tools; local coverage is suitable for sensitive services and machine-related connections.
Every MCP tool should differentiate between read-only and write-only. Actions such as production data access, deletion, permission modification, and external sending must be approved and audited. Tool output should be concise to avoid stuffing large amounts of sensitive or irrelevant data into context.
Study Checklist
Put the content on this page into real tasks and use the five dimensions of entry, context, permissions, verification and team rules to check whether you have truly mastered it.
Study Checklist
After reading this page, do not just remember the concept name. You should be able to place "Claude Code and MCP" back into a real Claude Code engineering workflow: where the task starts, what context the system loads, which actions need approval, how the result is verified, and how to roll back when it fails.
If this is a configuration or reference page, be specific about where the configuration is placed, whether it will be submitted, whether it contains sensitive information, whether it will extend the default permissions, and how to troubleshoot the settings that actually took effect in the event of a failure.
- Be able to describe in your own words the specific problem this page solves, rather than just reciting the title.
- Able to write a minimal example task with goals, scope, prohibitions, and acceptance criteria.
- Be able to determine which information should be put into the current prompt and which should be captured as project rules or configurations.
- Be able to explain which long-term rules should go into CLAUDE.md, and which runtime behavior should be handled by settings, permissions, Hooks, Skills and MCP.
- Ability to check diffs, command output, test results, screenshots or PR notes after a task is completed instead of just trusting the natural language summary.
If this page is used for team training, ask learners to complete a small task with Claude Code: read and explain first, submit a plan, make the smallest useful change, and close with real verification commands plus human diff review.