20 Best Memory MCP Servers, Compared (November 2026)

The short answer

Which one, for your situation

Top MCP servers for Memory

The ranking, with the evidence

• 1

  basic-memory

  Community favourite

  basicmachines-co/basic-memory ↗

  Best for: Obsidian users who want AI conversations to build a persistent Markdown knowledge graph: it exposes graph traversal, search, and note read/write tools over local files.

  Basic Memory exposes 12 MCP tools that create, read, edit, search, move, delete, list, validate, and follow relations between Markdown notes, plus project and manual management. For a local install it requires Python via uv and the --prerelease=allow flag; a hosted cloud option is available.

  Score 97.8: show every number behind it

  • Adoption100 / 100 · weight 40%
    • GitHub stars91
    • PyPI downloads82
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum100 / 100 · weight 14%
    • Stars gained, relative to size100
    • Stars gained, absolute100
    • PyPI download trend100
  • Tool quality83 / 100 · weight 13%
    • Tool description quality75
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  97.8

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  97.8

• 2

  codebase-memory-mcp

  Community favourite

  DeusData/codebase-memory-mcp ↗

  Best for: Navigating an unfamiliar multi-service repository: 17 tools index it into a local graph for symbol search, call-path tracing, and architecture queries, with no hosted service.

  codebase-memory-mcp exposes 17 MCP tools that index a repository into a persistent local knowledge graph and then answer questions through search_graph, query_graph (read-only Cypher), trace_path, get_file_outline, get_architecture, and index and coverage management tools. No npm or PyPI package is published, so installation comes from the single static binary in GitHub releases, and both source analysis and graph storage run on the local machine.

  Score 94.5: show every number behind it

  • Adoption100 / 100 · weight 40%
    • GitHub stars100
    • npm downloadsnot measured
      no npm package
    • Used through Glama14
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum93 / 100 · weight 14%
    • Stars gained, relative to size88
    • Stars gained, absolute100
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality66 / 100 · weight 13%
    • Tool description quality57
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  94.5

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  94.5

• 3

  MCP Memory Service

  Community favourite

  doobidoo/mcp-memory-service ↗

  Best for: Persisting and retrieving Claude memory across sessions: it exposes store_memory, retrieve_memory, and search_by_tag for semantic storage and tag searches.

  MCP Memory Service exposes three tools (store_memory, retrieve_memory, and search_by_tag) for storing tagged information, finding relevant memories by query, and searching by tags, and its README documents MCP and REST API interfaces. Before choosing it, note that it is a self-hosted PyPI package whose embeddings run locally via ONNX, so the service must be installed and run by the operator.

  Score 86.2: show every number behind it

  • Adoption88 / 100 · weight 40%
    • GitHub stars83
    • PyPI downloads33
      PyPI downloads show no weekday rhythm; halved
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum82 / 100 · weight 14%
    • Stars gained, relative to size85
    • Stars gained, absolute80
    • PyPI download trend79
  • Tool quality50 / 100 · weight 13%
    • Tool description quality50
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  86.2

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  86.2

• 4

  memory-bridge

  Community favourite

  jiabaobei/memory-bridge ↗

  Best for: Continuing an AI conversation across phone and PC: it synchronizes semantic memories with encrypted delta sync and preloads high-heat memories before the switch.

  memory-bridge is a Python MCP server that provides tools for adding, searching, and preloading semantic memories, and its README documents a membridge command-line interface for hybrid retrieval, automatic injection, handoff cards, sync, and a password-protected HTTP gateway for mobile devices. Before choosing it, note the shared-channel setup: cross-device synchronization uses a network-disk channel, with WebDAV or OneDrive configuration and credential handling documented for headless clients.

  Score 81.7: show every number behind it

  • Adoption66 / 100 · weight 40%
    • GitHub stars66
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum97 / 100 · weight 14%
    • Stars gained, relative to size100
    • Stars gained, absolute92
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality≈68 / 100 · weight 13%
    • Tool description quality≈68
      tool descriptions not yet scored
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  81.7

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  81.7

• 5

  iai-personal-memory-engine

  Community favourite

  CodeAbra/iai-personal-memory-engine ↗

  Best for: Recall of past decisions and preferences during a coding session: memory_recall and claim_check let an agent check prior rationale before searching the repository.

  It exposes 15 tools spanning verbatim capture and recall (memory_capture, memory_recall, memory_temporal_recall), store mutations for contradiction and reinforcement (memory_contradict, memory_reinforce), and read-only diagnostics such as topology, schema_list and events_query, alongside a CLI for daemon and hook installation. Automatic turn capture and context injection depend on hooks the host exposes, so a plain MCP-over-stdio client gets the tools only, and the documented platform support is macOS and Linux with Windows in beta.

  Score 81.2: show every number behind it

  • Adoption74 / 100 · weight 40%
    • GitHub stars74
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum56 / 100 · weight 14%
    • Stars gained, relative to size57
    • Stars gained, absolute54
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality83 / 100 · weight 13%
    • Tool description quality75
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  81.2

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  81.2

• 6

  TradeMemory Protocol

  Community favourite

  mnemox-ai/tradememory-protocol ↗

  Best for: AI trading agents that need persistent, outcome-weighted memory of past trades: it stores trades and recalls them scored by outcome quality, context similarity, and recency.

  TradeMemory Protocol exposes 20 MCP tools for recording trades in multi-layer memory, recalling outcome-weighted memories, reviewing strategy and behavioral performance, managing prospective trading plans, evolving strategies from Binance OHLCV data, and exporting and verifying audit records. The optional broker proxy requires Python 3.12+ and currently supports Alpaca.

  Score 80.1: show every number behind it

  • Adoption79 / 100 · weight 40%
    • GitHub stars79
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance95 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade80
    • Commit cadence100
  • Momentum42 / 100 · weight 14%
    • Stars gained, relative to size39
    • Stars gained, absolute48
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality83 / 100 · weight 13%
    • Tool description quality75
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  80.1

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  80.1

• 7

  marm-memory

  Community favourite

  Lyellr88/marm-memory ↗

  Best for: Developers switching between AI clients who need shared memory: it exposes 16 MCP tools for session logs, semantic recall, summaries, notebooks, and code/concept graph lookup.

  marm-memory is an MCP server that exposes 16 tools for session logging, semantic and keyword recall, notebook entries, summaries, compaction, distillation, and code/concept graph indexing and lookup. It assumes Python 3.10 or newer and, for shared memory across multiple agents, the swarm profile at startup.

  Score 78.2: show every number behind it

  • Adoption66 / 100 · weight 40%
    • GitHub stars66
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum58 / 100 · weight 14%
    • Stars gained, relative to size65
    • Stars gained, absolute48
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality83 / 100 · weight 13%
    • Tool description quality75
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  78.2

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  78.2

• 8

  Mneme Memory

  Emerging

  slow-stack/mneme ↗

  Best for: AI coding agents that need cross-session memory: it stores and semantically searches local SQLite memories with a Markdown mirror and human-reviewed conflict resolution.

  Mneme Memory provides cross-session memory for AI coding agents through save, search, list, get, update, and delete operations over a local-first SQLite store with a human-readable Markdown mirror, semantic search, background consolidation, and human-reviewed conflict resolution. Before choosing it, note that it requires the dsh-mneme DSH plugin running locally.

  Score 77.7: show every number behind it

  • Adoption58 / 100 · weight 40%
    • GitHub stars58
    • npm downloadsnot measured
      npm publishes mneme from onokumus/mneme
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum91 / 100 · weight 14%
    • Stars gained, relative to size100
    • Stars gained, absolute77
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality≈68 / 100 · weight 13%
    • Tool description quality≈68
      tool descriptions not yet scored
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  77.7

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  77.7

• 9

  Omega

  Community favourite

  omega-memory/omega-memory ↗

  Best for: AI agents used across multiple MCP clients need persistent local memory: it registers an MCP server that stores and recalls decisions and preferences across sessions.

  OMEGA documents an MCP server registration plus omega setup commands for Claude Desktop, Cursor, Windsurf, Cline, Codex, and Antigravity, with Claude Code hooks for automatic capture and surfacing and a Python API for store and query. Before choosing it, note that it requires Python 3.11 or newer and the [server] extra for MCP tools and hooks, while automatic capture and surfacing hooks are available only with Claude Code.

  Score 77.3: show every number behind it

  • Adoption70 / 100 · weight 40%
    • GitHub stars59
    • PyPI downloads61
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum53 / 100 · weight 14%
    • Stars gained, relative to size44
    • Stars gained, absolute29
    • PyPI download trend100
  • Tool quality≈68 / 100 · weight 13%
    • Tool description quality≈68
      tool descriptions not yet scored
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  77.3

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  77.3

• 10

  Shodh-Memory

  Community favourite

  varun29ankuS/shodh-memory ↗

  Best for: Offline agent and robot memory where no LLM calls are allowed: it stores and recalls persistent memory locally through MCP, HTTP, and ROS2/Zenoh interfaces.

  Shodh-Memory provides persistent memory for AI agents and robots through MCP, HTTP, and ROS2/Zenoh interfaces, with storage, recall, and forgetting handled locally and without an LLM in the loop. Its MCP tool list has not been inspected, and no npm package is published, so installation should use the PyPI package rather than the README's npx command.

  Score 76.3: show every number behind it

  • Adoption63 / 100 · weight 40%
    • GitHub stars63
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum68 / 100 · weight 14%
    • Stars gained, relative to size79
    • Stars gained, absolute51
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality≈68 / 100 · weight 13%
    • Tool description quality≈68
      tool descriptions not yet scored
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  76.3

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  76.3

• 11

  neural-memory

  Community favourite

  nhadaututtheky/neural-memory ↗

  Best for: Persisting AI agent memories across sessions without embedding API costs: it exposes remember/recall tools and recalls via graph-traversal spreading activation with explicit relationship types.

  Neural-memory exposes 10 MCP tools for storing memories, recalling them through spreading activation, and managing session or project context. It requires Python 3.11 or later and is installed from PyPI as neural-memory; no npm package is published.

  Score 75.1: show every number behind it

  • Adoption63 / 100 · weight 40%
    • GitHub stars60
    • PyPI downloads23
      PyPI downloads show no weekday rhythm; halved
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance97 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence85
  • Momentum57 / 100 · weight 14%
    • Stars gained, relative to size49
    • Stars gained, absolute33
    • PyPI download trend100
  • Tool quality73 / 100 · weight 13%
    • Tool description quality65
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  75.1

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  75.1

• 12

  PyMemoryEditor

  Community favourite

  JeanExtreme002/PyMemoryEditor ↗

  Best for: Reverse-engineering a running game or program with changing values: its scan, refine, pointer-chain and read/write tools operate on process memory.

  PyMemoryEditor provides an MCP server with 14 tools for opening processes, listing memory regions, scanning values or byte patterns, refining scan results, resolving pointer chains, and reading or writing addresses. Before choosing it, note that the MCP server is installed via the PyMemoryEditor[mcp] extra and launched with the pymemoryeditor-mcp command.

  Score 74.1: show every number behind it

  • Adoption65 / 100 · weight 40%
    • GitHub stars59
    • PyPI downloads42
      PyPI downloads show no weekday rhythm; halved
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance95 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade80
    • Commit cadence100
  • Momentum32 / 100 · weight 14%
    • Stars gained, relative to size36
    • Stars gained, absolute24
    • PyPI download trend34
  • Tool quality91 / 100 · weight 13%
    • Tool description quality83
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integrates100

  Weighted mean of the five

  74.1

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  74.1

• 13

  Graphify

  Community favourite

  Graphify-Labs/graphify ↗

  Best for: Giving AI coding agents codebase context without grepping: it builds a queryable knowledge graph over MCP, with fully local tree-sitter parsing for code.

  Graphify maps a project into a knowledge graph that AI coding agents can query over MCP, and its README documents invoking /graphify in an assistant to build that graph from code, docs, PDFs, images, and video. Code parsing runs locally with tree-sitter, but semantic passes for docs, PDFs, images, and video rely on the assistant's model or a configured API key.

  Score 72.0: show every number behind it

  • Adoption100 / 100 · weight 40%
    • GitHub stars100
    • PyPI downloads100
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum75 / 100 · weight 14%
    • Stars gained, relative to sizenot measured
      no snapshot history yet
    • Stars gained, absolutenot measured
      no snapshot history yet
    • PyPI download trend75
  • Tool quality≈68 / 100 · weight 13%
    • Tool description quality≈68
      tool descriptions not yet scored
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  92.3

  × relevance: the keyword is declared here

  0.78

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  72.0

• 14

  vshulcz/deja-vu

  Community favourite

  vshulcz/deja-vu ↗

  Best for: Recovering prior context from Claude Code, Codex CLI and opencode sessions: a single deja tool searches past sessions already saved to disk.

  deja-vu indexes the session histories that Claude Code, Codex CLI and opencode already write locally and exposes one MCP tool, deja, whose modes search those past sessions. Its recall is scoped to this user's own earlier sessions on the machine rather than general knowledge or library documentation, and syncing that store to another machine assumes SSH access between them.

  Score 71.6: show every number behind it

  • Adoption83 / 100 · weight 40%
    • GitHub stars79
    • npm downloads30
      downloads show none of the weekday rhythm human traffic has; halved
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum100 / 100 · weight 14%
    • Stars gained, relative to size100
    • Stars gained, absolute100
    • npm download trend100
  • Tool quality88 / 100 · weight 13%
    • Tool description quality80
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  91.8

  × relevance: the keyword is declared here

  0.78

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  71.6

• 15

  open-ontologies

  Community favourite

  fabio-rovai/open-ontologies ↗

  Best for: Ontology engineers changing production knowledge graphs: it exposes Terraform-style plan and apply tools plus DL checks to preview consequences before committing.

  Open Ontologies is a Rust MCP server exposing 42 tools for loading, validating, querying, diffing, reasoning over, and versioning OWL/RDF ontologies through an Oxigraph triple store. It assumes local setup for some features: embeddings require running open-ontologies init, crosswalk lookups require data/crosswalks.parquet, and the ontology store is in-memory.

  Score 71.0: show every number behind it

  • Adoption86 / 100 · weight 40%
    • GitHub stars86
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum100 / 100 · weight 14%
    • Stars gained, relative to size100
    • Stars gained, absolute100
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality73 / 100 · weight 13%
    • Tool description quality65
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  91.0

  × relevance: the keyword is declared here

  0.78

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  71.0

• 16

  MemoryGraph

  Community favourite

  memory-graph/memory-graph ↗

  Best for: Coding agents that can run shell commands and need persistent memory: it documents CLI commands to store typed memories, link them, and recall related context.

  The README documents a command-line memory interface for coding agents, with commands to store typed records, search or recall them, create relationships between records, and retrieve related memories. Before choosing it, note that setup runs from the repository with Bun and requires the memorygraph command to be linked and on PATH.

  Score 70.8: show every number behind it

  • Adoption60 / 100 · weight 40%
    • GitHub stars60
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance90 / 100 · weight 24%
    • Last commit touching this server88
    • Repository maintenance grade100
    • Commit cadence85
  • Momentum52 / 100 · weight 14%
    • Stars gained, relative to size60
    • Stars gained, absolute40
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality≈68 / 100 · weight 13%
    • Tool description quality≈68
      tool descriptions not yet scored
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  70.8

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  70.8

• 17

  Gortex

  Community favourite

  zzet/gortex ↗

  Best for: When indexing several repositories for an AI coding agent: it builds one cross-repo code graph and exposes it through CLI, MCP server, and web UI.

  Gortex indexes a code base into a knowledge graph and documents a CLI, an MCP server, and a web UI, with multi-repository support by default. It is installed as a single static binary for macOS, Linux, or Windows; no npm or PyPI package is listed.

  Score 68.7: show every number behind it

  • Adoption83 / 100 · weight 40%
    • GitHub stars83
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum95 / 100 · weight 14%
    • Stars gained, relative to size100
    • Stars gained, absolute87
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality≈68 / 100 · weight 13%
    • Tool description quality≈68
      tool descriptions not yet scored
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  88.1

  × relevance: the keyword is declared here

  0.78

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  68.7

• 18

  mcp-memory

  Steady

  fellowgeek/mcp-memory ↗

  Best for: Project-scoped agent memory across sessions: it exposes store, retrieve, search, delete, and session checkpoint tools backed by OKF markdown and SQLite indexing.

  The server exposes six MCP tools for storing, retrieving, searching, deleting, and checkpointing memory records, with records kept as OKF markdown files and indexed in SQLite. Every exposed tool requires an absolute project_root path, so the user must supply the active project root on each call.

  Score 68.7: show every number behind it

  • Adoption59 / 100 · weight 40%
    • GitHub stars59
    • npm downloadsnot measured
      no npm package
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance81 / 100 · weight 24%
    • Last commit touching this server88
    • Repository maintenance grade80
    • Commit cadence65
  • Momentum49 / 100 · weight 14%
    • Stars gained, relative to size57
    • Stars gained, absolute37
    • npm download trendnot measured
      no download history for the selected registry
  • Tool quality76 / 100 · weight 13%
    • Tool description quality68
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  68.7

  × relevance: the keyword is dedicated here

  1.00

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  68.7

• 19

  memorix

  Community favourite

  AVIDS2/memorix ↗

  Best for: Developers switching among multiple MCP-capable coding agents who need shared project memory: it exposes store, search, project context, context pack, session start, and detail tools.

  Memorix exposes nine MCP tools for storing and searching observations, building project context and context packs, checking CodeGraph status, resolving memories, starting sessions, fetching details, and managing local media. It requires an MCP-capable coding agent and keeps project memory under the Git project, with SQLite as the canonical local store.

  Score 68.5: show every number behind it

  • Adoption83 / 100 · weight 40%
    • GitHub stars74
    • npm downloads64
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum79 / 100 · weight 14%
    • Stars gained, relative to size100
    • Stars gained, absolute77
    • npm download trend42
  • Tool quality81 / 100 · weight 13%
    • Tool description quality73
    • Built and inspected by Glama100
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  87.8

  × relevance: the keyword is declared here

  0.78

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  68.5

• 20

  OpenLore

  Community favourite

  clay-good/OpenLore ↗

  Best for: AI coding agents working across refactors: it builds a local knowledge graph and exposes orient(task) to return relevant functions, callers, specs, tests, and insertion points.

  OpenLore wires an MCP server into a coding agent and documents memory and governance interfaces such as orient(task), certify_public_surface, change_impact_certificate, verify_claim, and openlore enforce. Its installation requires Node >=22.22 and npm, and it runs locally without an API key.

  Score 66.7: show every number behind it

  • Adoption78 / 100 · weight 40%
    • GitHub stars64
    • npm downloads68
    • Used through Glamanot measured
      not used through Glama in the last 30 days
  • Maintenance100 / 100 · weight 24%
    • Last commit touching this server100
    • Repository maintenance grade100
    • Commit cadence100
  • Momentum90 / 100 · weight 14%
    • Stars gained, relative to size100
    • Stars gained, absolute70
    • npm download trend98
  • Tool quality≈68 / 100 · weight 13%
    • Tool description quality≈68
      tool descriptions not yet scored
    • Built and inspected by Glamanot measured
      never built and inspected by Glama
  • Trust100 / 100 · weight 9%
    • License100
    • Published by the vendor it integratesnot measured
      not published by the vendor it integrates

  Weighted mean of the five

  85.5

  × relevance: the keyword is declared here

  0.78

  × continuity: actively changing

  1.00

  × evidence: widely adopted

  1.00

  Composite score

  66.7

Questions people ask

Other comparisons

• Best Figma MCP servers
• Best Obsidian MCP servers
• Best YouTube MCP servers
• Best Jira MCP servers
• Best PostgreSQL MCP servers
• Best MySQL MCP servers
• Best GitHub MCP servers
• Best Notion MCP servers
• Best Linear MCP servers
• Best Discord MCP servers
• Best MongoDB MCP servers
• Best CRM MCP servers
• Best Payments MCP servers
• Best Ecommerce MCP servers
• Best Social Media MCP servers
• Best Browser Automation MCP servers
• Best Email MCP servers
• Best PDF MCP servers
• Best Kubernetes MCP servers
• Best Docker MCP servers
• Best AWS MCP servers
• Best Spreadsheets MCP servers
• Best SEO MCP servers
• Best Web Search MCP servers
• Best Web Scraping MCP servers
• Best Vector Database MCP servers
• Best Code Execution MCP servers
• Best Calendar MCP servers
• Best Observability MCP servers
• Best Image Generation MCP servers
• Best Speech & Transcription MCP servers
• Best Maps MCP servers
• Best Travel MCP servers
• Best Stock Market MCP servers
• Best Web Analytics MCP servers
• Best Documentation MCP servers
• Best Security MCP servers