What information should a coding agent persist between steps?

Instruction: Explain what state a coding agent should carry forward during a task.

Context: Checks whether the candidate can explain the core concept clearly and connect it to real production decisions. Explain what state a coding agent should carry forward during a task.

Updated

Example Answer

I'd persist the task's acceptance criteria, repository commit, changed-file references, unresolved assumptions, and the commands and results already observed. I'd also record rejected hypotheses with a short reason, so the next step does not repeat the same failed approach.

For a paused retry bug fix, a useful checkpoint says which test reproduced the duplicate write and which patch is awaiting review. It should not say “all tests passed” if only one command ran. On resumption I'd compare the checkout with the recorded commit and refresh affected evidence. State needs access controls and retention limits because paths and logs can contain private information. A concise checkpoint is easier to verify than an entire unfiltered conversation.

Why this works

Defines observable workflow state, freshness checks, and the difference between evidence and an optimistic summary.

Practice follow-up

Which checkpoint fields must be refreshed after another engineer changes the branch?

Your preparation path

Work through these questions in order. Read the answer aloud, then explain it in your own words.

1. Start with the foundations

Build the vocabulary and explain the core decisions.

  1. When is a coding agent the right tool instead of autocomplete or manual engineering? Free sample
  2. What information should a coding agent persist between steps? Free sample
  3. Why do tests matter differently for coding agents than for chatbots? Free sample

2. Apply it to a real workflow

Practice diagnosis, validation, and everyday tradeoffs.

  1. What makes a patch proposal safe enough to apply automatically? Free sample
  2. The agent can find the bug, but its patches are hard for humans to review. What would you change? Member answer
  3. How would you verify a coding agent's web feature beyond a passing build and a screenshot? Free sample

3. Prepare for senior discussions

Explain failure boundaries, recovery, and production choices.

  1. A coding agent can modify infrastructure, app code, and data access layers in one task. How do you constrain it? Member answer
  2. Two coding agents change an API and its client in parallel. How would you integrate their work safely? Member answer
  3. A coding-agent hook runs before the repository trust prompt. What is wrong with that design? Member answer

Related Questions