Retrieve The Forgotten Memory
@roomi-fields
Retrieve The Forgotten Memory について
The open retrieval layer for AI coding agents. Indexes code, docs, legal, research, data — 15 parsers, FTS5 + semantic search, knowledge graph. Serves surgical context via MCP. Open source, local, free.
基本情報
設定
以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。
{
"mcpServers": {
"rtfm": {
"command": "rtfm-serve",
"args": [],
"env": {
"RTFM_DB": ".rtfm/library.db"
}
}
}
}ツール
ツールは検出されませんでした
ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。
概要
What is Retrieve The Forgotten Memory?
Retrieve The Forgotten Memory (RTFM) is a local, free knowledge-base indexer for AI agents. It indexes all project files—code, specs, PRs, PDFs, regulations, vault notes—into a single SQLite file and provides full-text, semantic, or hybrid search so agents find the right context before resorting to grep.
How to use Retrieve The Forgotten Memory?
Install with pip install rtfm-ai, then run rtfm init inside your project directory. Agents (e.g., Claude Code) query the indexed knowledge base automatically, seeing a 300-token metadata snippet before expanding only relevant content.
Key features of Retrieve The Forgotten Memory
- Indexes code, docs, PDFs, and any other file type
- Full-text, semantic, and hybrid search modes
- Progressive disclosure: 300-token metadata snippets
- 100% local—no cloud, no API keys, no cost
- Single SQLite file per project
- Runs in ~30 seconds to index a typical project
Use cases of Retrieve The Forgotten Memory
- AI coding agents that need access to project documentation, not just source code
- Legal or regulatory research across thousands of files and cross-references
- Large monorepos where standard grep searches miss the right file
- Keeping agent context efficient by expanding only relevant content
- Offline-first knowledge retrieval for sensitive or air-gapped projects
FAQ from Retrieve The Forgotten Memory
What makes RTFM different from code indexers like Augment, Sourcegraph, or Cursor?
Code indexers only see code. RTFM indexes everything—specs, PRs, architecture decisions, research papers, PDFs, regulations, vault notes—so the agent can find non-code context it needs to stop guessing.
What are the runtime dependencies and installation requirements?
RTFM requires Python. Installation is a single pip install rtfm-ai command with no API keys, cloud accounts, or external services.
Where does the indexed data live?
All data is stored in a single local SQLite file inside your project directory. Nothing is sent to any server.
What search methods does RTFM support?
It supports full-text search, semantic search, and hybrid search (combining both).
Does RTFM require internet or any authentication?
No. It runs entirely locally with no API keys, no cloud dependency, and no internet connection needed.
「メモリとナレッジ」の他のコンテンツ
Obsidian MCP Server
StevenStavrakisA simple MCP server for Obsidian
Semantic Scholar MCP Server
YUZongminA FastMCP server implementation for the Semantic Scholar API, providing comprehensive access to academic paper data, author information, and citation networks.
Jupyter Notebook MCP Server (for Cursor)
jbenoModel Context Protocol (MCP) server designed to allow AI agents within Cursor to interact with Jupyter Notebook (.ipynb) files
Notion MCP Server
makenotionOfficial Notion MCP Server
Rust Docs MCP Server
Govcraft🦀 Prevents outdated Rust code suggestions from AI assistants. This MCP server fetches current crate docs, uses embeddings/LLMs, and provides accurate context via a tool call.
コメント