Stage 7 · Lesson 5 of 5
MCP servers and background agents
The (MCP) is an open standard for connecting agents to outside tools and data, such as databases, issue trackers and browsers. Each server you add gives the agent new tools, and new ways to go wrong, so install servers only from sources you trust, prefer read-only access, and approve tool calls deliberately.
agents work without you watching: claude -p in a script, 's Cloud Agents, or Copilot's cloud agent, which takes an issue, works on a branch in a environment and opens a . They suit well-specified, well-tested chores. Because nobody watched them work, their output needs the most careful review of all.
Common trapmerging a background agent's pull request because CI is green, without reading the diff.
A question to check yourself, an AI practice session and 2 cheat sheet entries come when you add the roadmap to your account. Add it
Resources for this stage
Start with
- GuideFreeBuilding Effective Agents
Anthropic · Advanced
Anthropic's clear, free essay on how agents work and when simpler workflows are better
Keep handy
Also worth your time
Auto-review, Allowlist and Run Everything, permissions.json and sandboxing
- OWASP Top 10 for LLM ApplicationsGo deeper
Prompt injection and excessive agency explained by the security community