How do you give AI coding agents your product's context?

Give the agent the approved plan for the feature, with the decisions and business rules behind it, in a form it reads before it writes any code. In Rivora Connect, approved work goes to coding agents such as Claude Code, Cursor and Codex, with the decisions and rules attached, and the agent can check the open questions before it builds.

What context does a coding agent need besides the code?

The code tells an agent what exists. It doesn't say what your team decided. For a feature on an established product, the agent also needs:

• The approved plan: what to build, part by part, as your team agreed it.
• The business rules: the ones that apply across the product, and the ones for one client or market.
• The decisions and their reasons: so it doesn't undo something on purpose.
• What is still open: the questions nobody has answered yet, so it asks instead of assuming.

This is product context. It sits with your team, and Rivora keeps it as your product's memory.

How does it work with Claude Code, Cursor and Codex?

1. Your team designs and approves the feature in Rivora. A person approves each part, sends it back with a comment, or edits it.
2. The approved version is frozen as a handoff. It holds the approved parts, the business rules and the sources each part used.
3. The engineer connects the agent. People connect from Claude, Cursor or Codex by signing in with Rivora. No token is copied.
4. The agent reads the handoff and builds. It finds the feature by name, and Rivora asks it to read every approved part and check the open questions first. If something in the plan looks wrong, it can send a suggestion back for a person to review, if you allowed it when you signed in.

Is a CLAUDE.md or Cursor rules file enough?

For many teams, a context file is the right first step, and our free context file builder helps you write one. It's the best place for conventions, set-up and how your team likes code written.

Use Rivora alongside it when:

• several people and several agents work on the same product;
• rules differ by client, market or product area;
• decisions change, and the file needs to keep up;
• you want to see who confirmed each rule, and why.

The context file says how to work. Rivora says what to build, and why, from what your team approved.

What can the agent see?

• Approved handoffs in the workspace the person chose. Drafts never leave Rivora.
• The business rules, the sources each part used, and the open questions on the feature.
• Each connection can be seen and ended in Rivora, under Settings, Integrations, Agent access.

Questions

Does the agent need its own Rivora account?

The engineer signs in with their Rivora account from the agent, and picks the workspace. Rivora is in early access, and Rivora Connect sets up every team directly.

Does every engineer's agent see the same plan?

Yes. Each engineer signs in from the agent they use, and every agent reads the same approved version of the feature. An approved part changes only when someone chooses to change it.

Do we have to use all three agents?

No. Use the ones your engineers already use. Approved parts also export as Markdown or as a PRD for any other tool.