renor-engine

Renor Engine provides shared memory functionality for coding agents, enabling seamless recall of decisions and insights across different sessions. Designed for use with various coding environments, Renor allows agents to learn and share information collaboratively, enhancing productivity and reducing redundancy in communication.

How It Works

• Capture Knowledge Once: Share decisions, corrections, or important facts with an agent, which are then saved as personal Markdown files.
• Access Across Sessions: Any compatible agent, such as Claude Code, Codex, or Grok, can recall this knowledge in subsequent sessions, ensuring that all agents have access to the most up-to-date information.

Quickstart

Renor can be integrated easily in two primary ways:

Hosted Integration

With a simple command, users can connect Renor to their preferred agent.

npx renor connect --client <agent-name>

Replace <agent-name> with claude-code, codex, cursor, grok, etc. This method streamlines the setup process, including automatic account login and knowledge sharing features.

Self-hosted Option

Users can opt for a self-hosted solution to keep their information private. This setup requires Node.js version 22.16 or higher, and enables memory management locally.

claude mcp add --scope user renor-local -- npx -y renor-engine mcp

This approach ensures that created notes are stored locally as Markdown files that can be backed up, edited, and managed easily.

Memory Management

Renor operates with five core rules that guide agents on how to capture significant decisions and facts. This ensures that learned knowledge persists between agent sessions, enhancing overall knowledge sharing efficiency.

## Renor Memory Rule
- Before making decisions, always recall past knowledge using a get_context call.
- Save relevant information when a user corrects or provides important facts.
- Capture the exact value, avoiding summaries.
- Do not save sensitive information or data already present in the source code.

Features

• Markdown Storage: All learned knowledge is stored as plain Markdown files, easily accessible and editable.
• Searchability: The internal SQLite database provides fast search capabilities based on keywords and meanings, making it easy to retrieve past decisions and facts.
• Telemetry and Analytics: Renor includes options for telemetry to monitor usage anonymously, allowing for performance improvements without compromising user privacy.

Performance

Renor has shown impressive results in various benchmark tests, achieving high accuracy in recalling memories and insights gathered over multiple sessions. The engine is lightweight, with a simple installation and minimal resource requirements.

Learn More

Extensive documentation is available for users wishing to dive deeper into functionalities, self-hosting options, and system configurations: Documentation.

Renor Engine is positioned to become a valuable tool for developers seeking to enhance the collaborative potential of coding agents, ensuring that every learned detail is retained for future use.