Agent Engineering - Q&A Review Bank
30 questions across the module: harnesses, loops and graphs, context, coding and computer-use agents, skills and memory, verifiers and environments, and spec-driven development. Try each one before revealing the answer.
Learning objectives 45 min
By the end of this page you will be able to:- Recall the main design principles of agent engineering and the evidence behind them
- Apply them to short design scenarios
- Explain the trade-offs behind each principle
Prerequisites
- Chapters 1-8 of this module
Harness, Loops and Graphs
Check yourself
0 / 9 answered
- What is an agent harness?
- How do long-running coding harnesses carry work across context windows?
- Why separate generation from evaluation?
- What is a sprint contract?
- When should a harness component be removed?
- When does a recurring task justify building a loop?
- What limits how much a loop can improve an agent's output?
- Name three signals that justify turning a loop into a graph.
- What is the 'dies halfway' test?
Context
Check yourself
0 / 4 answered
- What is context rot?
- Just-in-time vs up-front context?
- Name five techniques for managing a long-running agent's context.
- How should compaction be scheduled given prompt caching?
Coding and Computer-Use Agents
Check yourself
0 / 5 answered
- What makes a repository agent-ready?
- What did METR's 2025 randomised trial of AI tools find?
- Interactive vs background coding agents?
- API, DOM automation or pixels - how do you choose for a computer-use task?
- Why are browser agents with logged-in sessions high risk?
Skills and Memory
Check yourself
0 / 5 answered
- What is in a skill, and which fields are required?
- Explain progressive disclosure for skills.
- AGENTS.md, skill, MCP server or memory - where does each kind of knowledge go?
- Why must skills be reviewed like code?
- Name the five policies of a memory subsystem.
Verifiers, Environments and Specs
Check yourself
0 / 7 answered
- Rank verifier types by reliability.
- What are the parts of an agent environment?
- How did RLVR extend to agents?
- Give three reward-hacking behaviours and a defence for each.
- What did Baker et al. find about chain-of-thought monitoring?
- Vibe coding vs vibe engineering?
- What are the stages of the Spec Kit workflow?
Last reviewed: 2026-09