Skip to main content
Glama
Sign In
eneszh
Turbo Quant Memory MCP Server

Turbo Quant Memory MCP Server

by Lexus2016
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Demo

Related MCP server: Sugar

The problem

A long session accumulates hard-won detail about why the code is the way it is. Then the context compacts and it is gone. Next session the agent re-reads the same files, re-derives the same conclusions, and bills you for the same tokens again.

CLAUDE.md does not scale past a few dozen lines, and it cannot answer "what did we decide about X, and why?".

Turbo Quant Memory is an MCP server that gives the agent a persistent, searchable store it writes to while it works — decisions, lessons, patterns, session handoffs — plus a compact index of your Markdown. Retrieval returns ~220-character result cards rather than whole documents; the agent loads full content only when a card is not enough.

Why this one

Turbo Quant Memory

mem0 / OpenMemory

MCP memory server

Where your data lives

your disk, always

vendor cloud or self-host

your disk

Your data leaves the host

never

yes, unless self-hosted

never

Retrieval

hybrid BM25 + dense vector, RRF-fused

dense vector

exact graph lookup

What a search returns

compact cards, hydrate on demand

full memories

full nodes

Knowledge graph

yes — with lifecycle + linting

no

yes

Non-English content

Cyrillic exact-match out of the box

varies

n/a

Measures its own savings

yes — server_info()

no

no

Price

free, MIT

paid tiers

free

No HTTP client, no telemetry, no phone-home. Verify it yourself — this returns nothing:

grep -rnE '^[[:space:]]*(import|from)[[:space:]]+(requests|httpx|aiohttp|urllib3)\b' src/

To be precise about the one exception: on first run fastembed downloads the embedding model (~0.22 GB) from Hugging Face. After that the server runs fully offline. Your notes, code and secrets are never transmitted anywhere — there is nothing in the package that could send them.

Install

Let your agent install it

Paste this into Claude Code, Codex, Gemini CLI, Cursor or Antigravity:

Install and configure the Turbo Quant Memory MCP server for this workspace from https://github.com/Lexus2016/turbo_quant_memory — follow the README, register the tqmemory server, run turbo-memory-mcp skill install, run the health check, and index this project.

skill install copies an operating manual into every agent skill directory on the machine, so every future session already knows how to use the memory without being told.

Or install it yourself

uv tool install turbo-quant-memory

Upgrading from 0.27.x or earlier? The distribution was renamed in 0.28.0, so uv tool upgrade turbo-memory-mcp no longer resolves — run uv tool install --force turbo-quant-memory once, and uv tool upgrade turbo-quant-memory afterwards. The turbo-memory-mcp command itself is unchanged, so client configs keep working.

Then register the server with your client:

claude mcp add --scope project tqmemory -- turbo-memory-mcp serve   # Claude Code
codex  mcp add tqmemory -- turbo-memory-mcp serve                   # Codex
gemini mcp add tqmemory turbo-memory-mcp serve                      # Gemini CLI

Cursor, OpenCode, Antigravity and other clients → CLIENT_INTEGRATIONS.md. Hermes runs MCP through a systemd gateway → docs/hermes.md.

📈 It measures its own savings — see for yourself

Turbo Quant Memory doesn't just claim to save tokens — every install keeps a running tally you can read anytime with server_info() (field usage_stats.headline). The savings are yours to verify, not ours to promise.

Live snapshot from a real developer instance (v0.28.2):

What the memory did

Number

🔢 Input tokens saved (cumulative)

≈ 2,640,000 and counting

🔁 Retrievals served

2,280 searches + 280 deep hydrations

📉 Average saved per retrieval

≈ 1,200 tokens

📚 Knowledge under management

237 active notes + 763 indexed code blocks

🛡️ Integrity

0 corrupted records · 0 pending migrations

These are one machine's cumulative numbers, not a synthetic benchmark — your own counter starts at zero and grows as your agent works. Run server_info() on your install to see your real figure.

What it does

  • Typed notes. decision, lesson, pattern, handoff — each stored with tags, provenance and a knowledge-graph link to the file or issue it is about.

  • Tiered memory. durable (decisions, patterns) and reference (indexed docs) are searched by default; episodic (session handoffs) stays out of the way until you ask for it, so yesterday's noise never buries an architectural decision.

  • Hybrid retrieval. A dense vector lane leads; a BM25 lane rescues exact terms — function names, file paths, IDs — fused with Reciprocal Rank Fusion. Cyrillic and other non-English terms match exactly, case- and accent-insensitive, with no configuration.

  • Knowledge graph. Directed, timestamped relations between notes, files and issues. Search results carry the linked context inline, so the agent does not need a second lookup.

  • Human notes outrank agent notes. Anything you explicitly asked to remember is flagged human-explicit and ranks above the agent's own observations at equal relevance.

  • Encrypted secrets vault. Project-scoped, AES-256-GCM, structurally unreachable from search. → docs/secrets-vault.md

  • Runs on a small machine. The default embedder is ONNX via fastembed — no PyTorch, ~0.22 GB model, comfortable on 2 GB of RAM.

  • Several agents, one project — automatically. Agents working in the same git working tree see each other's sessions, announced work and recent file changes; while two or more of them share the tree, changes to text files are recorded so they can be restored. No setup.

Full technical detail → docs/features.md

Several agents in one project

If you run more than one agent in the same working tree, tqmemory connects them automatically. There is nothing to configure.

  • Agents see each other. workspace_status lists the active sessions, what each one announced it is working on (announce_work, finish_work) and which files changed recently. A session that made no tqmemory call for 30 minutes is listed under idle_sessions instead: it is not an active peer and its announced areas are left out of overlaps. Other tqmemory responses carry a short workspace note only when something new has happened, and never for a lone session. health and server_info always include a workspace diagnostics block.

  • Changes are recorded. While two or more sessions share a git working tree, the watcher records new versions of text files it sees change (up to 1 MiB each, skipping anything matched by .gitignore or .tqmemoryignore). Several quick writes in a row can be merged into one version. file_history lists the versions and restore_file brings one back; the restore is itself undoable. The same is available from a terminal: turbo-memory-mcp files log and turbo-memory-mcp files restore.

  • Limits. Versions are recorded only while two or more sessions share a git working tree. The journal never runs outside a git working tree, in your home directory or at a filesystem root. Defaults: 1 MiB per file, a 256 MiB per-project size target and 14 days of retention. Recorded versions stay recoverable for 14 days; past that, or when the project is over its size target, the oldest are removed. The newest version of each file changed in the last 24 hours is never dropped for size. These defaults are set by TQMEMORY_JOURNAL_MAX_FILE_BYTES, TQMEMORY_JOURNAL_MAX_BYTES and TQMEMORY_JOURNAL_RETENTION_DAYS; the size limit is a target, not a hard maximum.

This is partial protection, not a guarantee that every change is saved. It does not block writes: two agents can still overwrite each other's work, and tqmemory can only show it and help you recover. Agents that never call tqmemory see nothing. It works on one computer only. To turn it off, set TQMEMORY_WORKSPACE_DISABLE=1.

Reuse one project's memory from another folder

For a temporary choice without a folder binding, call list_projects() and then select_project(project_id="<id>"). The current MCP connection uses that project's memory immediately. select_project() resets to automatic detection. Selection does not rewrite the project's original root or index local files. An explicit TQMEMORY_PROJECT_ID must not conflict with the selected ID.

list_projects() shows saved project IDs, names, display labels, original roots, and bound folders. The terminal equivalent is turbo-memory-mcp projects. To use an existing project's memory from a new folder, call bind_project(project_id="<id>", path="/absolute/new/folder"). The binding persists locally, and the current MCP connection switches to that folder immediately. server_info() confirms the resolved project. set_project_labels(project_id, labels=[...]) changes display labels; labels never determine identity. unbind_project(path="/absolute/new/folder") removes the exact binding.

Only existing project IDs can be bound. A binding does not index files in the new folder; call index_paths separately when you want that. Bindings are exact folders, so a nested repository is not silently captured. If a folder is replaced or its recorded Git origin changes, access fails until you explicitly rebind it. An explicit TQMEMORY_PROJECT_ID environment override still takes precedence.

Memory calls follow the selected folder; workspace_status, file_history, and restore_file still describe the MCP process's original working tree.

When restoring a storage snapshot on another machine, project notes, global notes, labels, and encrypted secrets move with the storage root. Local folder bindings contain the old machine's absolute paths; use select_project by ID and rebind folders on the new machine when needed. The secrets vault also requires its original passphrase or keyring key.

Automatic index maintenance

The primary storage-owning MCP process checks retrieval indexes immediately after startup and every six hours while it runs. It optimizes tables that have unindexed full-text rows or versions older than seven days. Work runs in the background under each project's in-process write lock; requests for that project may briefly receive a busy response while optimization holds the lock. Independent standalone or CLI writers are not covered by this lock; an optimization conflict is logged and retried on a later pass. Proxy clients and uncoordinated standalone fallback processes do not run maintenance. The source Markdown and canonical JSON notes are not changed. turbo-memory-mcp index-status reports table size, version age, and full-text coverage without modifying data.

LanceDB keeps recent versions for seven days, so optimization can temporarily increase disk use on an active table. Do not delete LanceDB files manually. If an older MCP process still owns the storage, restart its clients so the new process can become the owner and start maintenance.

The first full sync of a legacy table re-embeds its rows once. Later full syncs skip unchanged rows while the embedding model, backend, and backend version remain the same. Changing any of these rebuilds the vectors on the next full sync; scheduled maintenance alone does not re-embed them.

The 29 MCP tools

Group

Tools

Write

remember_note · deprecate_note · promote_note · index_paths

Read

semantic_search · hydrate · recent_context · list_scopes · list_projects

Projects

select_project · set_project_labels · bind_project · unbind_project

Graph

link_entities · unlink_entities · get_related_entities

Hygiene

lint_knowledge_base · health · self_test · server_info

Vault

set_secret · get_secret · list_secrets · delete_secret

Workspace

announce_work · finish_work · workspace_status · file_history · restore_file

Documentation

MEMORY_STRATEGY.md

How to actually use the memory day to day

CLIENT_INTEGRATIONS.md

Per-client setup: Cursor, OpenCode, Antigravity, …

TECHNICAL_SPEC.md

Architecture and storage format

docs/features.md

Retrieval, graph, tiers, embedder, FTS language

docs/secrets-vault.md

Vault setup, threat model, FAQ

docs/hermes.md

Hermes gateway setup and troubleshooting

CHANGELOG.md

Release history

License

MIT. Copy it, modify it, fork it, ship it inside a closed-source product, sell it. Attribution is the only condition.

Languages

🇺🇸 English · 🇺🇦 Українська · 🇷🇺 Русский

Available Tools

29 tools
announce_workC

Tell the other agents in this working tree what you are changing.

feature: short name of the work. areas: workspace-relative files or directories (globs allowed). Returns the peers whose announced areas overlap yours and recent changes inside them. Inferred writers are guesses from announced areas, never facts.

Inputs: feature (required), areas, details. Returns: object

bind_projectA

Bind a folder to an existing project and use it in this MCP session.

The folder must already exist. This does not index its files. Use the exact project_id from list_projects; labels are never identifiers.

Inputs: project_id (required), path. Returns: object

delete_secretC

Delete a project secret by exact name.

Inputs: name (required). Returns: object

deprecate_noteD

Inputs: note_id (required), scope, reason, replacement_scope, +1 more. Returns: object

file_historyA

Journaled versions of one file in this working tree, newest first.

Versions are snapshots taken while several sessions share the tree. Use restore_file with a version id to bring one back; the overwritten content is kept as a new version.

Inputs: path (required), limit. Returns: object

finish_workB

Clear the work this session announced with announce_work.

Inputs: no named parameters. Returns: object

get_related_entitiesA

Query relations involving a specific entity URI.

uri: the entity to look up, as note://<note_id>, file://<relative/path>, issue://KEY, task://KEY, or https://... — the same URI forms link_entities accepts. Returns nothing for an entity that has no links yet.

Inputs: uri (required), scope, relation_type. Returns: object

get_secretA

Fetch a project secret by exact name.

Returns the value in a dedicated secret_value field (never in descriptive text). Status is "ok" on hit, "missing" when no such name exists, or "error" with setup_hint when no master key is configured yet.

Inputs: name (required). Returns: object

healthD

Inputs: no named parameters. Returns: object

hydrateD

Inputs: scope (required), item_id (required), mode. Returns: object

index_pathsC

Register and index Markdown directories into project memory.

Supports .tqmemoryignore files (placed in project root or any indexed directory) with glob patterns to exclude paths from indexing. One pattern per line, # for comments. Example patterns: workspace-* skips any directory matching the glob; data/reports/*.md skips files matching a path pattern; backups/ (trailing slash, as in .gitignore) skips a directory. Agent worktrees under .claude/worktrees are skipped by default.

Inputs: mode, paths. Returns: object

link_entitiesC

Create a Knowledge Graph link between two entities.

Entities are specified using URIs:

  • Note: note://<note_id>

  • File: file://<relative_path> (relative to project root)

  • External: e.g. issue://BUG-404, task://TASK-101

Inputs: source_uri (required), target_uri (required), relation_type (required), scope. Returns: object

lint_knowledge_baseD

Inputs: paths, max_issues. Returns: object

list_projectsB

List saved projects with labels and explicitly bound folders.

Inputs: no named parameters. Returns: object

list_scopesD

Inputs: no named parameters. Returns: object

list_secretsA

List secret names in the active project. Never returns values.

Inputs: no named parameters. Returns: object

promote_noteD

Inputs: note_id (required). Returns: object

recent_contextA

Query-free session bootstrap: the most recently updated notes.

Call this FIRST when starting a new session or resuming after a context compaction, when you do not yet know what to search for. Returns notes ordered by recency (newest first), NOT by relevance — including handoff notes (episodic tier), which a plain semantic_search hides by default. This is the reliable "where did I leave off" entry point.

scope: 'project' (default), 'global', or 'hybrid'. Use 'hybrid' to also surface promoted cross-project knowledge. tier_filter: defaults to all tiers (so handoffs are included). Pass e.g. ["durable"] to exclude episodic session notes. format: 'json' (default); 'text' — compact plain text, one line per note with its hydrate handle (about 20% of the JSON size); 'hook' — the same text as Claude Code SessionStart hook output. Use 'hook' from a SessionStart mcp_tool hook: Claude Code shows the model only an mcp_tool hook's JSON additionalContext, never its plain text.

Inputs: limit, scope, format, tier_filter. Returns: object

remember_noteA

Store a typed project note. kind must be exactly one of: lesson, decision, pattern, handoff.

tags: 2-3 lowercase tags are strongly recommended — tags are the only way to FILTER notes; an untagged note is reachable by semantic_search alone. source_refs: the files / notes / issues this note is about. Pass them as entity URIs (note://, file://<relative/path>, issue://KEY, https://...) and they are auto-linked into the knowledge graph as references relations, so get_related_entities can later surface them. provenance: who originated the note. Use "human-explicit" when the USER asked to remember something ("remember this", "save that"); "agent" for your own observations (the default); "external-tool" for API/tool data; "system" for automatic processes. tier: normally derived from kind (handoff -> episodic, every other kind -> durable). Override e.g. tier="durable" to keep a handoff in the default-searchable set, or tier="episodic" to keep a noisy lesson out of regular search.

After storing, connect the note with link_entities(source_uri="note://", ...) when it relates to a file, issue, or prior note — the response echoes the note's uri and hints this when the note still has no relations.

Inputs: kind (required), title (required), content (required), tags, +4 more. Returns: object

restore_fileA

Write a journaled version of one file back into the working tree.

version: a version id from file_history, or "HEAD" for the committed content. The content being overwritten is kept as a new version, so the restore can be undone.

Inputs: path (required), version (required). Returns: object

select_projectB

Select an existing project by ID for this connection; omit ID to reset.

Вибирає наявний проєкт для цього підключення; без ідентифікатора скидає вибір.

Inputs: project_id. Returns: object

self_testD

Inputs: no named parameters. Returns: object

semantic_searchA

Compact memory retrieval (dense vector + BM25, fused via RRF).

By default only the durable and reference tiers are searched, so session handoff notes (which live in the episodic tier) are NOT returned. To recover handoffs / session summaries pass tier_filter=["episodic"] (or list every tier to opt everything in). For a query-free "where did I leave off" bootstrap at session start, prefer the recent_context tool instead.

source_filter narrows results by source type: "notes" returns only memory notes (decisions/lessons/patterns/handoffs), "markdown" only indexed doc blocks. In doc-heavy projects the reference blocks often crowd notes out of the top ranks on question-shaped queries — when you are asking "what did we decide/learn about X", pass source_filter="notes".

Inputs: query (required), limit, scope, tier_filter, +1 more. Returns: object

server_infoD

Inputs: no named parameters. Returns: object

set_project_labelsB

Set display-only labels for an existing project id.

Inputs: labels (required), project_id (required). Returns: object

set_secretA

Store an encrypted secret in the active project's vault.

The value is encrypted with AES-256-GCM under a per-project master key resolved from TQMEMORY_SECRETS_PASSPHRASE or the OS keyring. Secrets are NEVER indexed, embedded, or returned via semantic_search / hydrate. They live in ~/.turbo-quant-memory/projects/<project_id>/secrets/vault.tqv and stay on this machine.

Inputs: name (required), value (required). Returns: object

unbind_projectB

Remove an exact folder binding; reset this session if it used it.

Inputs: path. Returns: object

unlink_entitiesB

Remove a Knowledge Graph link between two entities.

source_uri / target_uri use the same URI forms as link_entities: note://<note_id>, file://<relative/path>, issue://KEY, https://...

Inputs: source_uri (required), target_uri (required), scope, relation_type. Returns: object

workspace_statusA

Overview of the sessions sharing this working tree: peers, overlaps and recent file changes.

Call it at session start. Inferred writers are guesses from announced areas, never facts. Peers that made no tqmemory call for a long time are listed under idle_sessions, not in sessions, and their announced areas are left out of overlaps.

Inputs: no named parameters. Returns: object

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Deploy Server
    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    TDQS

    C2.5/5.0

    Scored across 29 tools

    Disambiguation3/5

    Core clusters (notes, secrets, knowledge-graph links, multi-agent work) are reasonably distinct and the richer descriptions clarify boundaries like semantic_search vs recent_context. However, several tools have no description at all (self_test, health, server_info, list_scopes, hydrate, promote_note, deprecate_note, lint_knowledge_base), and the diagnostics trio plus list_scopes/list_projects and select_project/bind_project invite confusion. select_project's description is partially non-English, further reducing clarity.

    Naming Consistency4/5

    Names are overwhelmingly snake_case with a verb_noun or noun_state pattern (remember_note, link_entities, set_secret, list_secrets, index_paths, restore_file). Minor deviations are noun-only forms like health, hydrate, server_info, workspace_status, file_history, which are still readable and idiomatic. No camelCase/snake_case mixing.

    Tool Count3/5

    29 tools is heavy and spans five subdomains (notes, secrets, projects/indexing, multi-agent coordination, file journaling, plus diagnostics). Each cluster mostly earns its place, but the three overlapping diagnostics tools (self_test, health, server_info) look like padding that inflates the surface. It borders on too many for a 'memory' server.

    Completeness4/5

    Coverage is strong: full secret CRUD, note lifecycle (remember/promote/deprecate) plus retrieval and KG links, project bind/unbind/label/index, and coordination with file history/restore. Gaps exist, notably no update/edit or delete operation for notes and no explicit project creation tool (only binding to an existing project_id).

    Maintenance

    ActivityActive
    ResponsivenessLimited data

    Resources

    Related MCP Connectors

    Related MCP Servers

    • gemdex-mcp

      nikships
      A
      license
      Not graded
      quality
      B
      maintenance
      Provides AI coding agents with a persistent memory layer to save, recall, update, and track memories across sessions and repositories using hybrid semantic search.
      5
      MIT

    • Sugar

      Benjaminn581
      F
      license
      Not graded
      quality
      A
      maintenance
      Enables AI coding agents to read and write persistent, local-first memory across sessions and projects, with semantic search, global guidelines, and optional autonomous task execution.
      -

    • io.github.cdnsteve/sugar

      Sebastien94506
      F
      license
      Not graded
      quality
      A
      maintenance
      Gives AI coding agents persistent, local-first memory so they can store, semantically search, and recall decisions, preferences, error patterns, and cross-project guidelines directly during a session. It also optionally queues tasks for autonomous execution against that same memory.
      -

    • Sugar Memory

      Renjun22360
      F
      license
      Not graded
      quality
      A
      maintenance
      Provides AI coding agents with a local-first persistent memory layer they can read and write during sessions, storing decisions, preferences, error patterns, and research per project or globally. Enables semantic and keyword search, formatted context recall, and cross-project guideline surfacing without API keys or cloud storage.
      -