← Study Guide
4Exam Scenario

Developer Productivity with Claude

An agent that helps engineers explore unfamiliar codebases, understand legacy systems, generate boilerplate, and automate repetitive tasks using built-in tools (Read, Write, Bash, Grep, Glob) and MCP servers.

Tool Design & MCP IntegrationClaude Code Configuration & WorkflowsAgentic Architecture & Orchestration

What this scenario tests

Right built-in tool for the job

Grep for content search (callers, error strings, imports); Glob for name patterns; Read→Write fallback when Edit's anchor isn't unique. Build understanding incrementally rather than reading everything upfront.

MCP tool adoption

Enhance MCP tool descriptions so the agent doesn't default to built-in Grep over a more capable MCP tool. Expose content catalogs as MCP resources to cut exploratory calls.

Context management for large codebases

Delegate verbose exploration to subagents, keep scratchpad files, summarize phases before spawning the next, and design manifest-based crash recovery.

Edge Cases & Traps

  • Prefer community MCP servers for standard integrations; reserve custom servers for team-specific needs.
  • Watch for context degradation in long sessions (vague 'typical pattern' answers).

🧠 Knowledge Check · 15 questions

Score: 0/15

1.You need to find every caller of a function across the codebase. Which built-in tool?

2.The agent keeps defaulting to built-in Grep instead of a more capable custom MCP search tool. Best fix?

3.In a long exploration session, answers start referencing 'typical patterns' instead of the specific classes found earlier. What's happening and what helps?

4.For a standard Jira integration, what's the recommended choice?

5.You need to find every **/*.test.tsx file in the repo. Which built-in tool?

6.An Edit fails because its anchor text isn't unique. Reliable fallback?

7.Best way to build understanding of an unfamiliar codebase without exhausting context?

8.You want the agent to see the DB schema and issue catalog without repeated exploratory tool calls. What exposes that?

9.In a long exploration session, answers start citing 'typical patterns' instead of the specific classes found earlier. What's happening?

10.A verbose discovery phase is about to flood the main conversation. Best approach?

11.For a long multi-phase automation task, how do you make it resilient to a crash?

12.The agent keeps preferring built-in Grep over a more capable custom MCP search tool. Best fix?

13.Grep vs Glob — which does what?

14.You need to trace a function's usage across several wrapper modules. Efficient strategy?

15.For a standard Jira integration, what's recommended?