MnemoverseDocs

Appearance

ChatGPT Integration ​

Give ChatGPT persistent memory powered by Mnemoverse. Same memory as Claude Code, Cursor, and every other tool.

There are two ways to connect:

  • MCP connector (custom MCP server). ChatGPT connects to the hosted Mnemoverse MCP server and gets the memory tools. You sign in with your Mnemoverse account in the browser; there is no API key.
  • Custom GPT with Actions. A GPT calls the Mnemoverse REST API with your API key. Only Business, Enterprise and Edu workspaces can create GPTs, and OpenAI plans to retire custom GPTs.

Mnemoverse is not in the ChatGPT Plugin Directory

On 9 July 2026 OpenAI replaced the ChatGPT App Directory with the Plugin Directory, which ChatGPT and Codex share. Mnemoverse is not listed there, so ChatGPT has no Mnemoverse entry to install. You add the server yourself as a custom MCP server, as described below.

Sources, read 2026-10-01: ChatGPT release notes, entry for 9 July 2026: "We’re also replacing the App Directory with the Plugin Directory. Existing app connections are unaffected." Package your plugin: "Public plugins are published once to the universal plugin directory shared by ChatGPT and Codex."

Which path is open to you depends on your ChatGPT account

OpenAI opens custom MCP servers, developer mode and GPT creation to different plans:

Your planWhat is open to you
Free, GoNeither path. GPT creation is closed, and no current OpenAI page offers custom MCP servers or developer mode on these plans. Use Mnemoverse from Claude, Cursor or VS Code instead, or reach the same memory over the REST API.
Plus, ProA custom MCP server, but OpenAI's two pages disagree on writes, and neither names Plus. The help article says "Pro users can connect MCPs with read/fetch permissions in developer mode" and keeps write actions to Business and Enterprise/Edu. OpenAI's Add custom MCP server guide names no plan and says "ChatGPT supports both read and write tools from your server" (read 2026-10-11). Test memory_write before relying on it; recall works on either reading. On a Pro account, an existing Mnemoverse connector passed memory_stats, a write and a read (see MCP connector); sign-in and creating the server in ChatGPT were not run, and Plus has not been tested. If writes are refused, save from Claude, Cursor or VS Code or the REST API.
BusinessDeveloper mode for admins and owners only, with full MCP support, including write actions, which OpenAI describes as rolling out in beta. A member who is not an admin cannot turn developer mode on, and an admin cannot turn it on for them; members use the app once an admin publishes it to the workspace. Or the Custom GPT Actions path below with your mk_live_ key, where workspace permissions allow GPT creation.
Enterprise, EduDeveloper mode for admins and the members an admin grants it to, with full MCP support, including write actions, which OpenAI describes as rolling out in beta; or the Custom GPT Actions path below with your mk_live_ key.

Creating a Custom GPT is no longer possible on personal accounts. OpenAI's help centre is explicit: new GPT creation and publishing are not available on personal ChatGPT accounts, including Free, Go, Plus, and Pro. Creation lives in Business, Enterprise, and Edu workspaces, subject to workspace permissions. Existing GPTs keep running until OpenAI retires custom GPTs (dates below).

That is OpenAI's shape, not ours — the same limits apply to every memory service reached this way.

Checked against OpenAI's own guides on 2026-10-11. These limits move. OpenAI's help article has limited Pro connectors to read/fetch since at least October 2025, and GPT creation closed to personal plans in August 2026. If the table disagrees with what you see, OpenAI changed it and we have not caught up yet — tell us at support@mnemoverse.com and check their pages below, which are the authority here.

Sources: Creating and editing GPTs · Developer mode and MCP apps in ChatGPT · Add custom MCP server. OpenAI's earlier Developer mode guide, developers.openai.com/api/docs/guides/developer-mode, returns 404 as of 2026-10-11; Add custom MCP server now covers the same steps.

MCP connector (custom MCP server) ​

OpenAI changed this flow in October 2026

OpenAI replaced its developer-mode guide with an Add custom MCP server guide shortly before 2026-10-11, and its help article still describes developer mode. The two pages do not yet agree, and neither says which plans can add a custom MCP server. This section describes both as OpenAI states them on 2026-10-11. We re-check OpenAI's pages weekly (next: 2026-10-18) and will update this page as soon as the exact behaviour is clear. If you have connected Mnemoverse on a Plus, Pro, Business, Free or Go account, tell us what you saw at support@mnemoverse.com.

ChatGPT connects to the hosted Mnemoverse MCP server, added in ChatGPT as a custom MCP server: no Custom GPT, no API key, just a browser sign-in to your Mnemoverse account. OpenAI's Add custom MCP server guide says "Use ChatGPT on the web" and creates the server "as a plugin"; its help article calls the same thing a developer-mode app and answers "Are MCP apps available on mobile?" with "No - web only." (both read 2026-10-11).

Turn on developer mode ​

OpenAI's current Add custom MCP server guide has no developer-mode step: it goes straight to the steps in Add Mnemoverse as an app. OpenAI's help article still describes developer mode, per account:

  • Plus, Pro (personal): no current OpenAI page says where to turn developer mode on. The help article says "Pro users must continue enabling developer mode to use custom apps" and names no setting (read 2026-10-11). The Settings → Security and login path came from OpenAI's earlier developer-mode guide, which now returns 404. On our own Pro check of 2026-09-27 (see Source below), that page showed security controls but no Developer mode toggle, and chatgpt.com/plugins opened the Plugins settings. Go straight to the steps below. If Add custom MCP server is not offered, the same memory is reachable from Claude, Cursor or VS Code, or over the REST API.
  • Business: only admins and owners can use developer mode, and an admin cannot turn it on for a member. An admin turns it on under Settings → Apps → Advanced settings → Developer mode, or creates the app directly under Workspace settings → Apps → Create, then publishes it under Workspace settings → Apps → Drafts. Members find the published app in their Apps settings.
  • Enterprise, Edu: an admin first grants access under Permissions & Roles → Connected Data; each enabled user then turns it on under Settings → Apps → Advanced settings. Admins can also create the app under Workspace settings → Apps → Create.

OpenAI's Connect and test your plugin adds: "Account and workspace policies apply to adding and using custom MCP servers." (read 2026-10-11)

Add Mnemoverse as an app ​

  1. Open ChatGPT Plugins, select the + button, then Add custom MCP server. OpenAI's Plugins in ChatGPT article describes the same button differently, as opening "a conversation with Plugin Creator" (read 2026-10-11); if you get Plugin Creator and no Add custom MCP server, check the plan table above. In a workspace, an admin can also start from Workspace settings → Apps → Create.

  2. Enter a name and, optionally, a description, for example Mnemoverse.

  3. Under Connection, enter the server URL, including the /mcp path:

    text
    https://mcp.mnemoverse.com/mcp

    For authentication, choose OAuth. OpenAI lists "OAuth, No authentication, and OAuth or no authentication" (read 2026-10-11).

  4. Review the risk warning, select I understand and want to continue, then Create as a plugin, and sign in. The Mnemoverse server answers 401 Unauthorized until the OAuth sign-in is done, so ChatGPT cannot list the tools before you sign in. When ChatGPT opens the Mnemoverse sign-in page, sign in with your Mnemoverse account. The consent screen lists the access ChatGPT asked for, and you approve all of it or deny. If the list does not include Store and update your memories, ChatGPT will be able to recall but not save. In the workspace form the order is: click Scan Tools, complete the sign-in when it opens, wait for the scan to finish, then click Create.

  5. Review the tools ChatGPT found. There should be 10: every Mnemoverse tool except vault_list (see Not available in ChatGPT).

  6. Install the plugin. OpenAI's guide says to find it "in your personal plugins or the workspace where you created it" and to "Install it before using it in a conversation." (read 2026-10-11)

To use it, OpenAI's guide says: "In the prompt box, type @ and select your installed plugin." (read 2026-10-11). For an app an admin has published to the workspace, which members use without developer mode, OpenAI's help article says to "Start a chat and select one or more apps for a single message", and that the selection "applies to the message where you use it, not the entire conversation" (Use the new app in chats, read 2026-10-11). OpenAI's guide: "Write actions by default require confirmation." (read 2026-10-11). When the tool list changes on our side, open the app in ChatGPT Plugins and select Refresh to pull it. For an app published to a workspace, OpenAI's help article now says "Business and Enterprise/Edu admins can refresh an existing custom MCP app’s tool list from workspace app settings without recreating the app." (read 2026-10-11). The sign-in is OAuth; see Remote MCP Server for the flow and the tools it exposes.

These steps follow OpenAI's pages as read on 2026-10-11. We have not yet run them end to end on a new account; what we did run is in the Source note below.

Not available in ChatGPT ​

The connector gives ChatGPT every tool except vault_list. Mnemoverse does not offer Vault tools in ChatGPT, where access credentials are restricted data. vault_list works on other hosts, local and remote; Where each tool works lists every tool on every host.

Plan limits — verify before relying on writes

OpenAI's help article states, verbatim: "Apps, full MCP support, and developer mode are available for ChatGPT Business and Enterprise/Edu customers on ChatGPT web", with write actions "rolling out in beta" (Developer mode and MCP apps in ChatGPT, read 2026-10-11). OpenAI's Add custom MCP server guide names no plan and says instead "ChatGPT supports both read and write tools from your server" (read 2026-10-11). Plan behaviour changes faster than any docs page, ours included; test a write before relying on it.

Source, all read 2026-10-11. Add custom MCP server: "Use ChatGPT on the web.", "Select the plus button, then Add custom MCP server.", "Authentication options include OAuth, No authentication, and OAuth or no authentication.", "Select Create as a plugin.", "Install it before using it in a conversation.", "In the prompt box, type @ and select your installed plugin." and "Write actions by default require confirmation." Connect and test your plugin: "Go to ChatGPT Plugins.", "Enter a user-facing name and description.", "For a public endpoint, enter the MCP server URL, including the /mcp path.", "Review the tools and metadata discovered from the server." and "Account and workspace policies apply to adding and using custom MCP servers." Developer mode and MCP apps in ChatGPT: a direct fetch returned 403, so the article was read through the r.jina.ai reader: "Only admins/owners can enable developer mode and deploy an app. Admins cannot enable developer mode for individual members in their workspace.", "Business: Only Admins can use developer mode.", "Once published, apps appear in the workspace’s approved connectors list and in users’ Apps settings in ChatGPT", "Business and Enterprise/Edu admins can refresh an existing custom MCP app’s tool list from workspace app settings without recreating the app.", "Admins grant access in Permissions & Roles → Connected Data.", "After access is granted, enabled members can toggle developer mode for their account by navigating to Settings → Apps → Advanced Settings.", "Click Scan Tools and wait for the scan to complete. If your server uses OAuth, complete the authorization prompt, then wait for the tool scan to finish." and "Pro users must continue enabling developer mode to use custom apps." Plugins in ChatGPT, through the same reader: "go to Plugins and select Add (+) to open a conversation with Plugin Creator." OpenAI's earlier Developer mode guide (developers.openai.com/api/docs/guides/developer-mode), quoted here until 2026-10-11 for "Settings → Security and login" and for Plus eligibility, returns 404.

An existing Mnemoverse connector on ChatGPT Pro passed memory_stats, a write and a read on 2026-09-27; sign-in and creating the server in ChatGPT were not run. On that account, Settings → Security and login showed no Developer mode toggle, and chatgpt.com/plugins opened the Plugins settings.

Custom GPT with Actions ​

The alternative to the MCP connector: a Custom GPT that calls the Mnemoverse REST API through GPT Actions, with your API key. It gets the four actions under Available Actions, not the MCP tools. Creating a GPT needs a Business, Enterprise or Edu workspace.

OpenAI plans to retire custom GPTs

OpenAI's FAQ: "Custom GPTs are scheduled to retire on Dec 11, 2026." For affected Enterprise workspaces with an approved deferral, the date is Feb 11, 2027. In affected Enterprise workspaces, OpenAI also plans to end the creation of new custom GPTs on Oct 26, 2026; it notes that these dates are subject to change. The same FAQ says "GPT custom actions do not transfer through the migration workflow", so a GPT built with the steps below stops working at retirement. The custom MCP server above is not a GPT.

Source: Custom GPT retirement and migration FAQ, read 2026-10-01 through the r.jina.ai reader (a direct fetch returned 403).

How It Works ​

You → ChatGPT → GPT Action (HTTP) → core.mnemoverse.com
                     │
              X-Api-Key: mk_live_xxx

ChatGPT calls the Mnemoverse API directly through GPT Actions. No middleware, no proxy. Your GPT stores and recalls memories using the same API key as your other tools.

Setup (5 minutes) ​

Step 1. Get Your API Key ​

Sign up at console.mnemoverse.com and create a free API key. It starts with mk_live_.

If you already have a key from Claude Code or Cursor — use the same one. That's the point.

Step 2. Create a Custom GPT ​

Requires a Business, Enterprise, or Edu workspace

Personal accounts cannot create GPTs any more (see the note at the top). On a personal account, use the MCP connector above instead — it reaches the same memory.

  1. In chatgpt.com, signed in to a Business, Enterprise or Edu workspace, open Explore GPTs in the sidebar, or go to chatgpt.com/gpts.
  2. Select Create to open the GPT builder, and choose the configuration view to edit its fields directly.

These two steps follow OpenAI's Creating and editing GPTs, read 2026-10-01.

Step 3. Configure Actions ​

  1. Scroll down to Actions → Create new action
  2. Set Authentication:
    • Type: API Key
    • API Key: paste your mk_live_YOUR_KEY
    • Auth Type: Custom
    • Custom Header Name: X-Api-Key
  3. In the Schema box, click Import from URL and paste:
https://mnemoverse.com/docs/openapi-gpt.yaml

You should see 4 actions appear: writeMemory, readMemory, giveFeedback, getMemoryStats.

Step 4. Set the System Prompt ​

Paste this into the Instructions field (customize to your needs):

You have persistent long-term memory via Mnemoverse.

RULES:
1. BEFORE answering any question about preferences, past decisions, project 
   setup, people, or anything that might have been discussed before — call 
   readMemory first.
2. When the user shares a preference, makes a decision, teaches you something, 
   or tells you something important — call writeMemory immediately. Don't wait 
   to be asked.
3. After using memories to answer a question, call giveFeedback to report 
   whether they were helpful (1.0) or not (-1.0).
4. Your memory persists across sessions and across tools. What you learn here 
   is available in Claude Code, Cursor, and everywhere else.

WHAT TO REMEMBER:
- Preferences ("I prefer Railway over Heroku")
- Decisions ("We chose PostgreSQL for this project")
- Lessons ("Exponential backoff fixed the timeout issue")
- People ("Alice owns the design system")
- Project context ("Deploy target is staging.example.com")

WHAT NOT TO REMEMBER:
- Trivial small talk
- Temporary context ("open this file")
- Information already in the current conversation

Step 5. Test It ​

Say to your GPT:

"Remember that I always deploy to Railway using railway up"

The GPT calls writeMemory. Then start a new conversation and ask:

"How do I deploy?"

The GPT calls readMemory and recalls your preference. Same memory, new session.

Available Actions ​

Your GPT gets 4 actions:

ActionWhat it does
writeMemoryStore a preference, decision, or lesson
readMemorySearch memories by natural language query
giveFeedbackReport if a memory was helpful (+1) or wrong (-1)
getMemoryStatsCheck how many memories are stored

These are the same operations available in Claude Code, Cursor, and the Python SDK — just exposed through GPT Actions.

Tips ​

Make it proactive. The system prompt tells the GPT to store memories without being asked. This is key — users won't say "remember this", they'll just state preferences.

Use domains. If your GPT serves a specific project, add "domain": "project:acme" to writes and reads. This keeps memories organized.

Feedback matters. The giveFeedback action trains the memory system. Memories that get positive feedback rank higher in future searches. Over time, the most useful memories surface first.

Limitations ​

  • GPT Actions time out after 45 seconds round trip (Production notes on GPT Actions, read 2026-10-01).
  • ChatGPT may not call actions on every turn; the instructions from Step 4 guide it. Actions do not run in Pro mode: for a GPT with actions, ChatGPT offers only models that support them (Configuring actions in GPTs, read 2026-10-01).
  • Rate limits apply per organization, not per API key: all keys belonging to your organization share the same limits. Free tier: 1,000 queries/day. Upgrade for more.

OpenAPI Spec ​

The full spec is at /openapi-gpt.yaml. It defines 4 endpoints:

  • POST /memory/write — store a memory
  • POST /memory/read — search memories
  • POST /memory/feedback — rate usefulness
  • GET /memory/stats — memory statistics

This is a curated subset of the full API. For batch operations and advanced queries, use the REST API or Python SDK directly. Consolidation is not one of them: it is off on the hosted service, and POST /memory/consolidate answers 503 with a typed reason.

Cross-Tool Proof ​

This memory is shared with every other tool signed in to the same Mnemoverse account.

  1. In Claude Code: "Remember that the staging URL is staging.acme.com"
  2. In ChatGPT: "What's the staging URL?"
  3. ChatGPT calls memory_read and answers from what Claude Code saved.

Two things decide whether step 3 works

Your ChatGPT plan. This test saves in Claude Code and only asks ChatGPT to recall, so it does not depend on whether your plan can also save. The table at the top covers what OpenAI documents for each plan, including where OpenAI's pages disagree about Pro and say nothing about Plus.

Your Mnemoverse account. Memory belongs to the account, not to the tool or the key. Sign in to the ChatGPT connector with the same Mnemoverse account that Claude Code uses: the console account your API key was created in, or the account you signed in with. A Custom GPT needs a key from that account too. A different account opens a separate, empty memory, and nothing shows an error (why).

To check, ask both tools for their memory stats before step 2: memory_stats in Claude Code and in the ChatGPT connector, getMemoryStats in a Custom GPT. If ChatGPT reports zero memories and Claude Code does not, ChatGPT is signed in as a different account.

                    ┌── Claude Code (MCP)
                    ├── Cursor (MCP)
   Mnemoverse API ──├── VS Code (MCP)
   (one memory)     ├── ChatGPT (MCP / Actions) ← you are here
                    ├── Python SDK
                    └── REST API