"An exploration of deep search technologies or methods" matching MCP servers:

Enables semantic search over local knowledge bases like Obsidian notes, SQLite, and plaintext files, exposing results to Claude via MCP server.

AoT MCP server enables AI models to solve complex reasoning problems by decomposing them into independent, reusable atomic units of thought, featuring a powerful decomposition-contraction mechanism that allows for deep exploration of problem spaces while maintaining high confidence in conclusions.

Enables MCP clients to run offline-first web search, fetch and extract web content, build a local hybrid-retrieval knowledge base, and use persistent session memory without external API keys. It also supports intent-aware federated search, semantic reranking, entity graph exploration, and source-cited deep-search answers using local models.

A scientific reasoning framework that leverages graph structures and the Model Context Protocol (MCP) to process complex scientific queries through an Advanced Scientific Reasoning Graph-of-Thoughts (ASR-GoT) approach.

Enables semantic search across conversation archives via MCP, allowing AI clients to retrieve relevant past conversations using vector embeddings and text fallback.

A federation MCP server that sits in front of multiple memory backends and presents a unified search surface to AI agents, allowing a single query to search across knowledge graphs, session history, and web search.

Full-text search over Claude Code conversation history using SQLite FTS5, exposing indexed transcripts as MCP tools for searching, browsing, and reading turns.

An AI-focused search engine that enables AI applications to access high-quality knowledge from billions of webpages and ecosystem content sources across various domains including weather, news, encyclopedia, medical information, train tickets, and images.

An MCP server for managing contextual data as markdown files with metadata, enabling agents to save, retrieve, search, and delete contexts using simple CRUD operations.

Enables users to visualize the impact of edits in internal documents and emails by structuring content as a graph and tracing dependency chains.

Enables keyword search, filtering, and retrieval of Claude Code memory files, reducing context overhead by fetching only relevant memories on demand.

Enables semantic search over markdown files to find related notes by meaning rather than keywords, and automatically detect duplicate content before creating new notes.

MCP server for Yandex Wiki with full-text search. Read and write pages, comments, attachments, and dynamic tables (grids); optional server-side read-only mode for agents. Docker-ready.

MCP bridge for PDF Content Search — full-text PDF search with Apple Vision OCR across thousands of documents in under a second from Claude, Cursor, or any MCP client. Advanced filters (date, category, sender, amount), wildcards, boolean operators. Bridge open-source (MIT), PDF Content Search app is commercial with free iOS+Android companion scanner apps.

Persistent memory system for AI agents that records episodic memories with care-weighting and emotional valence, and provides full-text search with temporal chaining and automatic consolidation.

Enables fully local semantic search over a Roam Research graph via MCP, providing tools for hybrid retrieval, index refresh, and stats while keeping all content on-machine.

Enables deep research evidence synthesis and citation verification by generating structured research graphs through the Model Context Protocol.

A zero-config personal knowledge management MCP server that enables Claude Desktop to automatically classify, search, and organize local Markdown notes using hybrid search and Graph RAG.

Enables sophisticated reasoning workflows using graph-based representations for AI models.

Enables agents to run cross-runtime, cross-machine BM25 full-text search over historical conversations from Hermes, Google Antigravity, OpenClaw, Codex, and Claude Code, and to retrieve any past session's full transcript by ID. Queries return only ranked matching snippets, so recall costs no extra tokens or embeddings.