RepoMemory FAQs

Question 1

What kind of information can RepoMemory capture and recall?

Accepted Answer

It can capture git diffs, commit history, test outcomes (failures), terminal logs, and explicit engineering decisions. For recall, it offers file-level context, related-file discovery, recent failures, and decision/rationale lookup, all stored in a local SQLite database.

Question 2

What is RepoMemory and how does it help coding agents?

Accepted Answer

RepoMemory provides a repo-aware memory layer and MCP server for AI coding agents. It captures context from your coding sessions, such as failures, decisions, and file changes, and makes it available to subsequent agent sessions to prevent repeated mistakes and ensure continuity.

Question 3

Which AI coding agents and tools are compatible with RepoMemory?

Accepted Answer

RepoMemory is designed for cross-agent continuity, supporting popular AI coding agents like Claude, Codex, Cursor, and OpenHands. It integrates via its MCP (Model Context Protocol) server, making it compatible with any MCP-enabled client including VS Code and Claude Desktop.

Question 4

Is RepoMemory local or cloud-based for storing memory?

Accepted Answer

RepoMemory is a local-first tool. It stores all your memory entries, file relationships, and session data in a local SQLite database directly within your repository's `.repomemory/` directory, ensuring privacy and portability without relying on external cloud services.

Question 5

How does RepoMemory prevent repeated mistakes across agent sessions?

Accepted Answer

It records crucial information like failed tests, terminal logs, git diffs, and engineering decisions in a local memory graph. When a new agent session starts, it can recall this context, ensuring it's informed by past attempts and doesn't repeat previous errors.

RepoMemory

RepoMemory

主な機能

ユースケース

主な機能

ユースケース