Mission 9 of 12 · Updated November 2026

Build an AI Recruitment Agent with Tools & MCP

Go from one LLM call to an agent. Write your own JavaScript agent harness - a bounded loop of model, tool, result, model - with a tool registry, conversation state, execution limits, allowlists and human approval before anything consequential. Then connect it to an MCP server and trace every step in Langfuse.

By the end of this mission: “I can build a bounded, controllable AI agent with tool calling, MCP and human approval, and explain exactly how the loop works.”

The tech stack for this mission

OpenAI logoOpenAINode.js logoNode.jsZod logoZodMCP logoMCPExpress logoExpressPostgreSQL logoPostgreSQLLangfuse logoLangfuse

What you learn, checkpoint by checkpoint

  1. Checkpoint A

    Define tools with Zod schemas and run a bounded agent loop

    • The difference between LLM calls, workflows and agents
    • Define tools with Zod input schemas and use OpenAI tool calling
    • Write the agent loop yourself: model, tool, result, model, with a hard stop

  2. Checkpoint B

    Add state, limits, allowlists, retries and human approval

    • Harness engineering: a tool registry, conversation state and execution limits
    • Tool permissions and allowlists, plus human approval before consequential actions
    • Handle tool errors with retries and clear stopping conditions

  3. Checkpoint C

    Connect an MCP server and trace the whole agent in Langfuse

    • MCP fundamentals: clients, servers, tools and transports
    • Connect your agent to an MCP server over Streamable HTTP and debug it with MCP Inspector
    • Trace the agent and every tool call in Langfuse

AI companion mode · AI as Product

AI runs inside what you build

AI lives inside the product you build. Treat its output as untrusted - validate it, trace it, evaluate it, and design for uncertainty.

What proves you’re done

Working agent with 2+ tools, an MCP integration, a human-approval gate and Langfuse traces

Every mission ships something real you can show. If you can’t explain it, you can’t submit it.

Deliberately left out of this mission

Multi-agent frameworks · LangGraph · Autonomous coding sandboxes · Advanced memory · Complex orchestration

Depth is set on purpose: you master what this mission needs, not everything at once. The advanced material lives in optional self-study after Mission 12.

Free · no card needed

Reading is not building.

This mission is part of the paid 12-week programme - but the way we teach is free to try. The free trial puts you inside Sigmo, the learning portal our students use daily, with the 6 Projects in 6 Days precourse: real projects, real code, published to your own GitHub.

Start the free trial

Mission 8

Build an AI Recruitment Assistant with RAG

Mission 10

Make the Agent Reliable: Tracing, Evals & Safety

Ready for the real thing?

The AI-Native Software Development Programme - all 12 missions with a live cohort, daily Buildroom, and mentors watching your back.

Apply nowSee the whole curriculum