Lesson 2.4Lesson 11 of 30
MCP servers in Claude Code and agents
Integrate MCP servers into Claude Code and agent workflows
Key points
- 1Shared team servers go in the project .
mcp.json(committed). Personal or experimental ones go in the user's ~/.claude.json. - 2Keep secrets out of git with environment variable expansion: "${
GITHUB_TOKEN}" in .mcp.json. - 3Tools from all configured servers are discovered at connection and available at the same time.
- 4If the agent keeps using Grep instead of your MCP tool, improve the MCP tool's description. Expose catalogs as MCP resources so the agent doesn't explore with tool calls.
- 5For standard integrations (e.g. Jira) use an existing community server; build a custom one only for team-specific workflows.
What the exam guide checks here
Objective wording from the exam guide (English).
- 2.4.K1Scope: project .mcp.json for shared team tooling vs user ~/.claude.json for personal/experimental
- 2.4.K2Environment variable expansion in .mcp.json (${GITHUB_TOKEN}) keeps secrets out of git
- 2.4.K3Tools from all configured servers are discovered at connection and available simultaneously
- 2.4.K4MCP resources expose content catalogs to reduce exploratory tool calls
- 2.4.S1Configure shared servers in project .mcp.json with env var expansion
- 2.4.S2Configure personal servers in user ~/.claude.json
- 2.4.S3Improve MCP tool descriptions so the agent stops preferring built-ins like Grep
- 2.4.S4Prefer community MCP servers for standard integrations (Jira); custom only for team-specific workflows
- 2.4.S5Expose catalogs as MCP resources instead of exploratory calls
Lock it in
Answer scenario questions on this topic. Every option comes with an explanation.
Practise 2.4 · 23 questions