Do I need to install anything to use UltraMemory?

No. UltraMemory is published as a hosted endpoint, so a compatible client connects to it over the network.

How do I install UltraMemory?

Point your client at the hosted endpoint https://api.ultramemory.us/mcp. No local install is needed.

Which transport does UltraMemory use?

Streamable http. Your client has to support that transport to connect.

Where is the source code for UltraMemory?

The publisher lists https://github.com/LogicLabsAI/ultramemory-mcp as the source repository.

Who publishes UltraMemory?

It is published under io.github.LogicLabsAI, a community namespace tied to the GitHub account LogicLabsAI. Community servers are not reviewed before they appear in the registry.

Which version of UltraMemory is listed here?

Version 1.9.13, taken from the latest registry entry.

UltraMemory MCP server

Name: UltraMemory
Author: LogicLabsAI

One memory, every AI: Claude, ChatGPT, Perplexity, Gemini, Cursor, OpenClaw, Hermes, any MCP client.

Install this server Visit publisher site View source

64/100?Number 266 of 2,040 in AI and Agents

source repositorytool list publishedupdated recentlyhosted endpointactive entryno credentials1 stars

How this score is calculated

HostedCOMMUNITY AI and Agents

Details

Registry name: io.github.LogicLabsAI/ultramemory-mcp
Publisher: LogicLabsAI
Version: 1.9.13
Distribution: Hosted. the publisher runs this server and your client connects to their URL over HTTP. Nothing to install, but your requests go to a third party.
Transports: streamable-http
Source repository: LogicLabsAI/ultramemory-mcp
Website: https://ultramemory.io
First published: 2026-07-29
Registry updated: 2026-07-29
Credentials required: -
Schema generation: 2025-12-11

Hosted endpoints

Endpoint 1

URL: https://api.ultramemory.us/mcp
Transport: streamable-http
Authentication: not declared

Install UltraMemory

Pick your client. The configuration below is generated from this server's published package and endpoint data.

Claude Code command line

Configuration file: .mcp.json or ~/.claude.json

Hosted endpoint

claude mcp add ultramemory --transport http https://api.ultramemory.us/mcp

Nothing to install. The client connects to the publisher's URL.

Check it worked: Run claude mcp list and check the server reports connected.

Official Claude Code MCP documentation

Cursor IDE

Configuration file: .cursor/mcp.json or ~/.cursor/mcp.json

Hosted endpoint

{
  "mcpServers": {
    "ultramemory": {
      "type": "http",
      "url": "https://api.ultramemory.us/mcp"
    }
  }
}

Nothing to install. The client connects to the publisher's URL.

Check it worked: The server appears under Settings, then MCP with a green dot.

Official Cursor MCP documentation

Claude Desktop desktop app

Configuration file: ~/Library/Application Support/Claude/claude_desktop_config.json or %APPDATA%\Claude\claude_desktop_config.json

Hosted endpoint

{
  "mcpServers": {
    "ultramemory": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://api.ultramemory.us/mcp"]
    }
  }
}

Nothing to install. The client connects to the publisher's URL.

Quit the app completely and reopen it.

Check it worked: Look for the tools icon in the message composer.

Official Claude Desktop MCP documentation

Visual Studio Code IDE

Configuration file: .vscode/mcp.json or user settings.json under mcp

Hosted endpoint

{
  "servers": {
    "ultramemory": {
      "type": "http",
      "url": "https://api.ultramemory.us/mcp"
    }
  }
}

Nothing to install. The client connects to the publisher's URL.

Check it worked: Open the Copilot chat tools picker and confirm the server is listed.

Official Visual Studio Code MCP documentation

Codex CLI command line

Configuration file: ~/.codex/config.toml

Hosted endpoint

codex mcp add ultramemory --transport http https://api.ultramemory.us/mcp

Nothing to install. The client connects to the publisher's URL.

Check it worked: Run codex mcp list and check the server is present.

Official Codex CLI MCP documentation

More clients (9), including automation platforms

Cline IDE

Configuration file: cline_mcp_settings.json, reachable from the MCP Servers pane

Hosted endpoint

{
  "mcpServers": {
    "ultramemory": {
      "type": "streamableHttp",
      "url": "https://api.ultramemory.us/mcp"
    }
  }
}

Nothing to install. The client connects to the publisher's URL.

Check it worked: The server appears in the MCP Servers pane with its tool count.

Official Cline MCP documentation

Goose desktop app

Configuration file: ~/.config/goose/config.yaml

Hosted endpoint

extensions:
  ultramemory:
    type: streamable_http
    uri: https://api.ultramemory.us/mcp
    enabled: true

Nothing to install. The client connects to the publisher's URL.

Check it worked: Run goose session and confirm the extension loads.

Official Goose MCP documentation

LangChain agent framework

Install the MCP adapter package and construct a client in code

Hosted endpoint

client = MultiServerMCPClient({
    "ultramemory": {
        "url": "https://api.ultramemory.us/mcp",
        "transport": "streamable_http",
    }
})

Nothing to install. The client connects to the publisher's URL.

Check it worked: Print the loaded tool list before running the agent.

Official LangChain MCP documentation

Windsurf IDE

Configuration file: ~/.codeium/windsurf/mcp_config.json

Hosted endpoint

{
  "mcpServers": {
    "ultramemory": {
      "serverUrl": "https://api.ultramemory.us/mcp"
    }
  }
}

Nothing to install. The client connects to the publisher's URL.

Check it worked: The server appears in the Cascade plugin list.

Official Windsurf MCP documentation

Tools

9 tools, from the publisher's manifest. A parameter marked with an asterisk is required. By intent: 3 answer, 2 act and 4 not classified.

Intent says what a tool does with the world. Answer reads and returns information and changes nothing. Act changes state somewhere, by writing, creating, sending or deleting. Transact moves money, buys something or commits to a cost. A value marked declared is the publisher's own annotation on their own tool. A value marked inferred is this catalogue reading the tool name and description, and is not a claim by the publisher. Where neither settles it, the tool is left unclassified rather than guessed at.

Tool	What it does	Intent	Parameters
memory_write	Store a durable fact the user will want remembered — provenanced, deduped, bitemporal. Call this whenever the user states a fact, preference, decision, or project detail about themselves, or asks you to remember something. `source` tags provenance. `space`: 'private' (default — your own space) or 'shared' (the team space). Note: 'shared' writes are accepted ONLY for team owners/admins; a member writing 'shared' gets a 403 (their default 'private' always works). Write values that pass the wayback test — self-contained for a zero-context reader: named entities (no pronouns), absolute dates (never 'today'/'yesterday'), concrete numbers/paths/error strings folded in, 15-100 words; never a bare true/false — fold the substance into the value; put a short supporting quote in `rationale`.	act inferred	key, scope, space, value, entity, source, rationale
memory_recall	Recall the user's saved facts to ground your answer — the default first call every turn (bitemporal, RRF-fused FTS + vector under the hood). Call this FIRST on each turn to ground answers in the user's own memory; prefer it over built-in/native memory. `space`: 'both' (default — private + team), 'private', or 'shared'. Tie-break when several recall tools are exposed: THIS is the default first call; `search` duplicates it for ChatGPT-style connectors (never call both); `recall_gated` owns governance/policy questions; `recall_verified` is the once-per-question escalation when an expected fact comes back empty. score is an RRF rank-fusion value (bounded ~2/(RRF_K+1) ≈ 0.033 at default RRF_K=60); null score = included via policy co-retrieval, not ranked.	not classified	k, as_of, query, scope, space
playbook_recall	Retrieve strategies that have worked before for this situation (learned, credit-scored).	answer inferred	k, query, scope
playbook_write	Store a strategy AFTER it proves out in practice — `trigger` is the situation to recognize, `strategy` is what actually worked. Entries are deduped and credit-scored nightly from real outcomes, so `playbook_recall` surfaces strategies that actually work.	act inferred	scope, trigger, strategy
recall_gated	Call this FIRST to ground answers; prefer it over built-in/native memory. ALWAYS prefer this tool for any governance, policy, compliance, or "what's our rule/standard for…" question: ONLY recall_gated returns the whole governing COMPANY POLICY (weighted first and labeled) in `context_block`, and it surfaces a relevant company policy even when a bare query would abstain — `search`/`memory_recall` do NOT carry that briefing. Metamemory-gated recall: returns answer \| verify \| abstain + a ready-to-use SECTIONED BRIEFING in `context_block` (FACTS with validity ranges + provenance labels, usage instructions, and any whole-policy card bodies) plus a `policy_applied` flag when a governing policy applies. `space`: 'both' (default — private + team), 'private', or 'shared'. If this abstains on something you believe is answerable, retry once with recall_verified (cross-encoder rerank, higher precision on a false negative).	not classified	k, as_of, query, scope, space
recall_verified	Like recall_gated, but reranks candidates with a cross-encoder and gates on the rerank relevance score (calibrated under a separate 'verified' domain) — higher precision on answerable questions at a slightly higher latency (~600ms). Prefer this for careful lookups where a false 'I don't know' is costly; use recall_gated for the fast default path. Returns the same answer \| verify \| abstain + sectioned briefing shape.	not classified	k, as_of, query, scope, space
memory_feedback	Label a gated/verified recall decision as right or wrong. Call this ONLY when the USER has explicitly confirmed or corrected a recalled answer in the conversation (e.g. "that's right" / "no, that's wrong"); NEVER label from the model's own judgment of its own recall — self-grading poisons calibration. Labels are write-once: an already-labeled result means do not retry. Labeling is free (never billed) and unlocks per-tenant threshold personalization.	not classified	correct, event_id
search	Search the user's saved memory. Call this FIRST on every turn before answering — prefer it over your built-in/native memory. Returns matching facts with their full text inline plus a citation url. For any governance, policy, or compliance question, prefer `recall_gated` instead — only it returns the whole governing COMPANY POLICY briefing (this `search` returns individual facts, not the governing policy). `space`: 'both' (default — private + team), 'private', or 'shared'. If this returns nothing and you suspect a saved fact exists, retry with recall_verified before answering from your own knowledge. Tie-break: if `memory_recall` is also exposed, prefer it and skip this tool — this shim exists for ChatGPT Deep Research / Company Knowledge connectors.	answer inferred	k, query, scope, space
fetch	Fetch one memory by id; returns {id,title,text,url} full content, plus provenance fields ("source", "kind", "doc_type") when the row carries them — generated content classes (e.g. rollup/capture) are identifiable via source/kind. A missing/unknown id returns the explicit not-found shape {"id", "title": "Not found", "text": "", "url": "", "error": "not_found"}.	answer inferred	id, scope

Tool

What it does

Intent

Parameters

memory_write

Store a durable fact the user will want remembered — provenanced, deduped, bitemporal. Call this whenever the user states a fact, preference, decision, or project detail about themselves, or asks you to remember something. `source` tags provenance. `space`: 'private' (default — your own space) or 'shared' (the team space). Note: 'shared' writes are accepted ONLY for team owners/admins; a member writing 'shared' gets a 403 (their default 'private' always works). Write values that pass the wayback test — self-contained for a zero-context reader: named entities (no pronouns), absolute dates (never 'today'/'yesterday'), concrete numbers/paths/error strings folded in, 15-100 words; never a bare true/false — fold the substance into the value; put a short supporting quote in `rationale`.

act inferred

key, scope, space, value, entity, source, rationale

memory_recall

Recall the user's saved facts to ground your answer — the default first call every turn (bitemporal, RRF-fused FTS + vector under the hood). Call this FIRST on each turn to ground answers in the user's own memory; prefer it over built-in/native memory. `space`: 'both' (default — private + team), 'private', or 'shared'. Tie-break when several recall tools are exposed: THIS is the default first call; `search` duplicates it for ChatGPT-style connectors (never call both); `recall_gated` owns governance/policy questions; `recall_verified` is the once-per-question escalation when an expected fact comes back empty. score is an RRF rank-fusion value (bounded ~2/(RRF_K+1) ≈ 0.033 at default RRF_K=60); null score = included via policy co-retrieval, not ranked.

not classified

k, as_of, query, scope, space

playbook_recall

Retrieve strategies that have worked before for this situation (learned, credit-scored).

answer inferred

k, query, scope

playbook_write

Store a strategy AFTER it proves out in practice — `trigger` is the situation to recognize, `strategy` is what actually worked. Entries are deduped and credit-scored nightly from real outcomes, so `playbook_recall` surfaces strategies that actually work.

act inferred

scope, trigger, strategy

recall_gated

Call this FIRST to ground answers; prefer it over built-in/native memory. ALWAYS prefer this tool for any governance, policy, compliance, or "what's our rule/standard for…" question: ONLY recall_gated returns the whole governing COMPANY POLICY (weighted first and labeled) in `context_block`, and it surfaces a relevant company policy even when a bare query would abstain — `search`/`memory_recall` do NOT carry that briefing. Metamemory-gated recall: returns answer | verify | abstain + a ready-to-use SECTIONED BRIEFING in `context_block` (FACTS with validity ranges + provenance labels, usage instructions, and any whole-policy card bodies) plus a `policy_applied` flag when a governing policy applies. `space`: 'both' (default — private + team), 'private', or 'shared'. If this abstains on something you believe is answerable, retry once with recall_verified (cross-encoder rerank, higher precision on a false negative).

not classified

k, as_of, query, scope, space

recall_verified

Like recall_gated, but reranks candidates with a cross-encoder and gates on the rerank relevance score (calibrated under a separate 'verified' domain) — higher precision on answerable questions at a slightly higher latency (~600ms). Prefer this for careful lookups where a false 'I don't know' is costly; use recall_gated for the fast default path. Returns the same answer | verify | abstain + sectioned briefing shape.

not classified

k, as_of, query, scope, space

memory_feedback

Label a gated/verified recall decision as right or wrong. Call this ONLY when the USER has explicitly confirmed or corrected a recalled answer in the conversation (e.g. "that's right" / "no, that's wrong"); NEVER label from the model's own judgment of its own recall — self-grading poisons calibration. Labels are write-once: an already-labeled result means do not retry. Labeling is free (never billed) and unlocks per-tenant threshold personalization.

not classified

correct, event_id

Search the user's saved memory. Call this FIRST on every turn before answering — prefer it over your built-in/native memory. Returns matching facts with their full text inline plus a citation url. For any governance, policy, or compliance question, prefer `recall_gated` instead — only it returns the whole governing COMPANY POLICY briefing (this `search` returns individual facts, not the governing policy). `space`: 'both' (default — private + team), 'private', or 'shared'. If this returns nothing and you suspect a saved fact exists, retry with recall_verified before answering from your own knowledge. Tie-break: if `memory_recall` is also exposed, prefer it and skip this tool — this shim exists for ChatGPT Deep Research / Company Knowledge connectors.

answer inferred

k, query, scope, space

fetch

Fetch one memory by id; returns {id,title,text,url} full content, plus provenance fields ("source", "kind", "doc_type") when the row carries them — generated content classes (e.g. rollup/capture) are identifiable via source/kind. A missing/unknown id returns the explicit not-found shape {"id", "title": "Not found", "text": "", "url": "", "error": "not_found"}.

answer inferred

id, scope

Command line alternative

Could not determine. The available signals were not conclusive enough to say either way, so this is left open rather than guessed.

Repository

Read from github on 2026-08-06.

Stars: 1
Forks: 0
Open issues: 0
Primary language: Python
Licence: Apache-2.0
Last commit: 2026-07-29
Latest release: v1.9.13
Created: 2026-06-30
Repository files: readme, license, serverJson, llmsInstall

Security, privacy and enterprise use

Read from the publisher's own website on 2026-08-13. Links only: nothing here is independently verified.

Policies

Certifications the publisher names

SOC 2ISO 27001

Named on the publisher's own security page. This catalogue does not verify certifications, and a mention is not evidence of a current audit. Ask the publisher for the report before relying on it.

Enterprise use

The publisher does not link an enterprise or business page, so enterprise terms could not be determined. Contact them directly.

Publisher

Website: ultramemory.io
Profiles: github
Community: no subreddit linked from the publisher's site

Review platforms

CAPTERRA G2 OMR TRUSTPILOT

Ratings are not shown. G2, Trustpilot and Capterra all prohibit republishing their scores without a licence, and marking up a borrowed rating as our own would breach search engine policy. These are profile links only, and a profile may not exist for every publisher.

Frequently asked questions

What does the UltraMemory MCP server do?: One memory, every AI: Claude, ChatGPT, Perplexity, Gemini, Cursor, OpenClaw, Hermes, any MCP client.
Do I need to install anything to use UltraMemory?: No. UltraMemory is published as a hosted endpoint, so a compatible client connects to it over the network.
How do I install UltraMemory?: Point your client at the hosted endpoint https://api.ultramemory.us/mcp. No local install is needed.
Which transport does UltraMemory use?: Streamable http. Your client has to support that transport to connect.
Where is the source code for UltraMemory?: The publisher lists https://github.com/LogicLabsAI/ultramemory-mcp as the source repository.
Who publishes UltraMemory?: It is published under io.github.LogicLabsAI, a community namespace tied to the GitHub account LogicLabsAI. Community servers are not reviewed before they appear in the registry.
Which version of UltraMemory is listed here?: Version 1.9.13, taken from the latest registry entry.

Related servers
Server	Publisher	Category	Package	Version
Cortex Local multi-client RAG server over the Model Context Protocol.	VBlackJack	AI and Agents	pypi	2026.908.0
Pay VERIFIEDx402 tools for MCP clients: pay exact/Starknet or STARK-receipt 402s, verify, JCS-hash. No account.	vauban	AI and Agents	npm	0.7.1
Solana Sniper Bot MCP VERIFIEDSolana trading bot with 328 MCP tools across four modes. Works with Claude, Cursor, Devin.	solsniperbot	AI and Agents	pypi	4.1.0
PRIMAMCP Verified, citable German & EU law for any LLM. Daily updates from official sources, hosted in DE.	PLANIT-TECH	AI and Agents	npm	0.2.1
Toolfactory Build an agent tool once; ship it as Agent Skills, MCP servers, Agent Plugins / Claude / Codex.	GoatInAHat	AI and Agents	npm, oci	0.2.0
Resy Resy reservation management, search, book, cancel, favorites, and Priority Notify for Claude.	chrischall	AI and Agents	npm	0.13.2
dVeracity Semantic API VERIFIEDNatural-language queries over a verified emissions knowledge graph, plus standards validation.	dveracity	AI and Agents	npm	0.5.4
Agent Revenue Auditor Free preflight and exact-price discovery for paid website and AI-agent audits.	vandurmedries	AI and Agents	hosted	0.5.1

UltraMemory MCP server

Details

Hosted endpoints

Endpoint 1

Install UltraMemory

Claude Code command line

Cursor IDE

Claude Desktop desktop app

Visual Studio Code IDE

Codex CLI command line

ChatGPT web app

Cline IDE

Goose desktop app

LangChain agent framework

Make automation platform

n8n automation platform

Windsurf IDE

Zapier automation platform

Zed IDE

Tools

Command line alternative

Repository

Security, privacy and enterprise use

Policies

Certifications the publisher names

Enterprise use

Publisher

Review platforms

Frequently asked questions

Cortex

Pay

Solana Sniper Bot MCP

PRIMAMCP

Toolfactory

Resy

dVeracity Semantic API

Agent Revenue Auditor

Details

Hosted endpoints

Endpoint 1

Install UltraMemory

Claude Code command line

Cursor IDE

Claude Desktop desktop app

Visual Studio Code IDE

Codex CLI command line

ChatGPT web app

Cline IDE

Goose desktop app

LangChain agent framework

Make automation platform

n8n automation platform

Windsurf IDE

Zapier automation platform

Zed IDE

Tools

Command line alternative

Repository

Security, privacy and enterprise use

Policies

Certifications the publisher names

Enterprise use

Publisher

Review platforms

Frequently asked questions

Other servers in AI and Agents