Kerebrom FAQs

Question 1

What are Kerebrom's core memory features?

Accepted Answer

Kerebrom provides a 'seven-tool memory surface': 'context' for session awareness, 'recall' for specific queries, 'remember' for durable observations, 'summary' for consolidating work, 'timeline' for chronology, 'forget' for invalidating observations, and 'projects' for administrative management.

Question 2

Does Kerebrom require manual intervention to manage memory?

Accepted Answer

No, Kerebrom is designed to feel automatic. It features an 'auto setup' to configure memory hooks for detected AI agents. Agents then utilize semantic tools like 'context', 'recall', 'remember', and 'summary' proactively, so you don't have to explicitly prompt them to use memory.

Question 3

What is Kerebrom and what problem does it solve?

Accepted Answer

Kerebrom is a local-first persistent memory engine designed for users working with multiple AI assistants. It solves the problem of siloed AI memories by providing a unified, durable memory layer that agents like Claude, Codex, and Gemini can share, ensuring consistent context across all your AI interactions.

Question 4

How does Kerebrom ensure privacy and data security?

Accepted Answer

Kerebrom prioritizes privacy and security with a local-first model. All memory data is stored securely in your local SQLite + FTS5 database, with no telemetry, cloud database, or hosted accounts. Remote MCP access is strictly opt-in, giving you full control over your data.

Question 5

Which AI agents and clients does Kerebrom support?

Accepted Answer

Kerebrom is designed to integrate with a wide range of AI clients, including Claude Code, Claude Desktop Chat, Cowork, Codex, Cursor, Gemini CLI, OpenCode, Windsurf, and VS Code, providing a shared memory layer for these and any MCP-capable client.

Kerebrom

Kerebrom

Key Features

Use Cases

Key Features

Use Cases