Claude Code sometimes puts a suggested next message in the input box after finishing a task. Zohaib Ansari wondered whether the real benefit was not saving developers a few keystrokes, but learning from what they accept or change. He clearly called it a guess — and then updated the essay with a correction from someone on the Claude Code team.

  • Ansari’s theory: an edited suggestion, such as changing “run the tests” to “run only the auth tests,” could reveal what a developer who knows the project actually wants next.
  • His broader point: predicting the next request could help a coding agent learn the sequence of real work, from changing code to checking it.
  • The correction: Anthropic’s edwinarbus says these suggestions are not used to collect preference signals. They were built to help people stay in the flow or pick up a session later. The team measures how many suggestions are accepted to assess whether the feature helps; session training use depends on plan and privacy settings.

The 132-comment thread on Hacker News adds a useful challenge to the original theory: a prompt shown to you changes what you might say next.

What the thread adds

  • SubiculumCode — argues that displaying the guess “biases the user’s response toward the prediction, making it less independent.” That weakens the idea that an accepted suggestion is a clean preference signal.
  • bugos and tonmoy — independently ask why a model couldn’t predict the next message after a user writes it unprompted, rather than showing a suggestion first. A reply from ismailmaj offers a distinction: multiple different follow-ups could all be acceptable, so accepting a proposed one might reveal something that comparison with a single unprompted message cannot.
  • tripleee — reports a concrete failure mode: Claude changed a feature they did not ask to change, then suggested reverting it. A plausible next-step suggestion is no proof that the preceding work was sound.

HN handles are pseudonymous; HN publishes no per-comment scores. This is a slice of the thread in HN’s ranking order, not a consensus.

The better lesson is about control, not a secret data pipeline: suggested next steps can make coding agents easier to use, but they can also steer what gets checked — or skipped. The author’s update makes the distinction between a plausible incentive and an evidenced product decision unusually clear.