"Construct 3" matching MCP servers:

Enables persistent, auditable business operations state by storing knowledge, evidence, decisions, tasks, execution receipts, and review results in a local SQLite database, so different sessions can continue the same project.

An MCP server for executable mathematics that enables agents to construct objects, compute invariants, search for witnesses, and verify results with independently checkable evidence.

Read-only MCP access to the p-e relay store: an append-only corpus of provenance-bearing coordination records, with tools to fetch a record by id, list records and replies, and check existence. This image serves a snapshot frozen at build time; append_relay is listed but refused, and deposits to the live store require an HMAC-signed request.

Enables persistent, project-scoped memory for OpenAI Codex Desktop and CLI on Windows, capturing prompts and responses through managed hooks, with federated recall across projects and optional local graph extraction.

Provides local, scoped access to DEVONthink 3 on macOS, enabling search, reading cited text, following references, and previewing document organization from an AI assistant.

MARM MCP provides persistent memory and structured session context beneath any AI tool, so your agents learn, remember, and collaborate across all your workflows.

Description: Persistent memory for AI agents with rollback, audit trails, semantic search, and knowledge graph. Zero-config local SQLite or cloud API. 23 tools, 6 resources, 3 prompts.

The most feature-complete MCP server for Obsidian vaults. 23 tools and 3 resources for search, read, write, tags, link analysis, graph traversal, and canvas support.

MCP server for foldkit that exposes the 7-prime spine, 7 κ-bands, and 6 fold operations as native tools and resources in any MCP client.

Enables MCP clients to search a Zotero library, read items, take notes, organize collections, and perform semantic search over PDFs.

Enables persistent memory management with semantic vector recall, BM25 keyword search, and 3-way fused retrieval combining lexical, semantic, and tag-based signals.

Provides a compiled knowledge substrate, extracting atomic claims from an append-only capture log and querying a bitemporal claim graph over MCP. Read-only by default, with opt-in writes for trusted sources.

Persistent AI memory server with 3-layer hybrid search (vector + FTS5 + keyword), confidence scoring via Reciprocal Rank Fusion, episodic/profile memory, and 16 tools. Zero LLM dependency. Works standalone with Claude Desktop and Claude Code. MIT licensed.

A TypeScript Model Context Protocol (MCP) server to allow LLMs to programmatically construct mind maps to explore an idea space, with enforced "metacognitive" self-reflection.

Integrates with Get笔记 (GetBiji) API to enable AI-powered knowledge base search and retrieval. Provides tools for enhanced semantic search, raw content recall, and rate limit monitoring.

Enables AI agents to safely write to and read from persistent memory by validating every write through three independent consensus nodes, storing memories only on 2/3 agreement and quarantining poisoning attempts.

DescriptiShared knowledge cache for AI agents — cache-first search saves tokens and avoids redundant web searches. Cross-agent deduplication with trust scoring. Human Bridge for blocked/paywalled content. MCP-native (FastMCP), ChromaDB-backed. 3 tools: agenthive_search, agenthive_contribute, agenthive_stats.

Enables AI agents to store, search, list, and forget long-term memories in a fixed namespace backed by S3-compatible storage, with optional read-only mode and provenance-tagged recall results.

Persistent AI memory server with 3-layer hybrid search (vector + FTS5 + keyword), confidence scoring via Reciprocal Rank Fusion, episodic/profile memory, and 16 tools. Zero LLM dependency. Works standalone with Claude Desktop and Claude Code. MIT licensed.

Enables querying PDF documents using natural language with grounded answers and source citations via a local RAG pipeline.