Skip to main content

​
The problem

Claude Code persists context via CLAUDE.md files. But this mechanism has two limits:
  1. It doesn’t scale in size. CLAUDE.md is injected into the system prompt, which means it eats into the context window. The more rules you add, the less room the agent has to work. The Context Graph stores knowledge externally and the agent queries only what it needs.
  2. It’s siloed per engineer. When one engineer defines coding rules in their agent, the rest of the team never sees them. Scale that to 10 engineers running parallel agents, and you get 10 diverging sets of conventions. Each agent “remembers,” but none of them share.

​
How Rippletide Code fixes this

Rippletide Code gives Claude Code persistent, structured memory backed by a Context Graph. Instead of re-explaining your conventions every session, you store them once. Every agent session reads from the same source of truth. The graph stores your engineering standards:
  • Naming conventions — camelCase for variables, PascalCase for components, UPPER_SNAKE_CASE for constants
  • Architectural patterns — where to put hooks, how to structure API routes, which state management to use
  • Design system rules — approved components, spacing tokens, color usage
  • Error handling policies — how to handle API failures, logging standards, retry strategies
Rippletide Code process flow

​
What you get

Rippletide Code covers five areas:
CapabilityHow it works
SetupOne command to authenticate, scan your repo, select rules, and install hooks
Rule-guided codingRules are injected into every Claude session automatically. Violations are blocked before code is written.
Rule managementAdd, edit, or delete rules using natural language — no config files to maintain
Rule sharingShare your rule set with teammates via email and OTP
PlanningGenerate implementation plans that are automatically reviewed against your rules

​
Integration

Rippletide integrates with Claude Code through hooks — shell commands that run automatically at key points during the agent session. The connect command sets up these hooks so the Context Graph is queried seamlessly, without any manual intervention. Two hooks run during every coding session:
  • fetch-rules.sh (UserPromptSubmit) — fires on every prompt, queries the backend with the user’s request, and injects relevant rules into Claude’s context
  • check-code.sh (PreToolUse) — fires before every Edit/Write, sends the proposed code to the backend for violation checking, and blocks the tool call if violations are found

​
Data privacy

Rippletide only relies on the context available inside your local Claude Code workflow:
  • the current Claude Code chat session for the active project
  • your CLAUDE.md
Your codebase is not centrally analyzed outside of your local Claude Code environment. The analysis happens through your own Claude Code setup. Rippletide only stores the rules extracted from that context, not the rest of the project content. In practice, this means Rippletide keeps reusable conventions and policies, not your full codebase or unrelated session content.

​
Next steps

Connect

Set up Rippletide Code in your project

Coding Session

How rules are enforced while you code

Rule Management

Add, edit, and delete rules with natural language

Rule Sharing

Share rules with teammates

Planning

Generate rule-compliant implementation plans

Team Governance

Scale conventions across engineers and repositories