Product History

What product history means and how eve records it.

Product history is the explanation of what changed in the product and why the change can be trusted.

Git can tell you a file changed. eve should tell you what shipped, which behavior changed, how the work was verified, and what future contributors should know before building on it.

What belongs in product history

An eve snapshot should describe a finished unit of work from the reader's point of view:

  • the title and type of change
  • the product summary
  • the user-visible behavior, when there is one
  • decisions that explain why the result looks this way
  • risks, tradeoffs, or follow-up concerns
  • validation commands and outcomes
  • artifacts such as screenshots, videos, logs, URLs, and notes
  • relationships to earlier snapshots

A good snapshot is useful months later to someone who was not present when the work happened.

What does not belong

eve is not a replacement for commit messages, code review comments, issue trackers, or long-form design docs.

Avoid using snapshots for:

  • Every intermediate commit
  • Raw terminal dumps without interpretation
  • Private credentials or secrets
  • Unverified claims
  • Exhaustive implementation narration that belongs in code review

Keep the snapshot focused on the product result and the evidence for it.

What good looks like

A weak snapshot says "fixed auth bug."

A useful snapshot says "users are redirected back to the dashboard after GitHub OAuth; verified with go test ./... and a login-flow screenshot; risk remains around expired provider sessions."

The second version gives a future reader enough context to understand the shipped behavior and decide what to check before changing it.

How agents benefit

Agents are good at reading structured local context. eve gives them a stable source of truth that is closer to product intent than a commit log.

Before changing a feature, an agent can inspect related snapshots to understand prior decisions, risks, and validation expectations. After completing work, the agent can write a new snapshot that future agents and humans can trust.

On this page