Guide

How do I give Claude Code or Cursor persistent product context?

Prompts and scratch notes are ephemeral. A Product Map is persistent product context: what the product is, how it is structured, and what each feature is for — reusable across sessions and tools.

What “persistent” actually means

Persistent context survives a new chat, a new agent, and a new teammate. It is not a longer system prompt. It is a maintained model of domains, jobs, flows, capabilities, and features that agents can read and that humans can correct.

How Kriyastream fits Claude Code and Cursor

Keep the Product Map as the source of truth in Kriyastream. When you work in Claude Code or Cursor, point agents at the relevant map structure (and linked code evidence) instead of retyping product lore.

Pair this with Code Intelligence so features stay connected to repos and paths — agents implement against named product units, not vague tickets.

  • Product Map = durable what/why
  • Work breakdown = durable what to build next
  • Code links = durable where it lives

How to do it

  1. 1Create or import your Product Map in Kriyastream.
  2. 2Confirm features and descriptions so agents inherit accurate language.
  3. 3When starting a coding session, reference the map domains/features in scope.
  4. 4After shipping, align code evidence back to features so context stays current.

FAQ

Is this the same as a rules file or AGENTS.md?

Rules files guide agent behavior. A Product Map captures product structure. You want both: rules for how to work, Product Map for what the product is.

See also Claude Code and Cursor pages

Dedicated integration pages explain Kriyastream + Claude Code and Kriyastream + Cursor workflows.