@mnemoverse/mcp-memory-server
Installed, started with no arguments, no environment and no input, and listed its tools. Run 2026-11-16.
Needs to start: nothing
Returns a result when called with no setup: 0 of 4 tools tried
Contacts while installing and starting: nothing
Its tool definitions changed since version 0.17.2. If you approved the earlier ones, read them again.
What it contacted
Nothing besides the npm registry it was installed from. Watched from outside its sandbox while it was installed and started with no input. Its tools may still contact things when called.
Starting it
npx -y @mnemoverse/mcp-memory-server@0.18.1
Installing a package runs its code. Use a sandbox for anything you have not read.
Calling its tools
4 tools that take no input and look read-only were called once with nothing. 0 returned a result.
memory_list_recent | No usable answer tool reported an error: Mnemoverse: no API key is configured, so this tool cannot run. Tell the user to set MNEMOVERSE_API_KEY in their MCP client config — a fr |
memory_stats | No usable answer tool reported an error: Mnemoverse: no API key is configured, so this tool cannot run. Tell the user to set MNEMOVERSE_API_KEY in their MCP client config — a fr |
memory_list_rooms | No usable answer tool reported an error: Mnemoverse: no API key is configured, so this tool cannot run. Tell the user to set MNEMOVERSE_API_KEY in their MCP client config — a fr |
vault_list | No usable answer tool reported an error: Mnemoverse: no API key is configured, so this tool cannot run. Tell the user to set MNEMOVERSE_API_KEY in their MCP client config — a fr |
During the calls it contacted nothing it had not already contacted.
Only tools whose names read like looking something up, with no required input, are tried, five at most. Tools that need input, and anything that sounds like it changes or sends something, are not called.
What its tools tell an agent
In the tool descriptions, nothing that talks about the user, other tools or private files, or uses hidden characters.
1 tool with commands in capitals or very long text
- Gives the agent orders in
memory_list_recent:cursor is how you get it. To catch up on a shared room you MUST pass its address as `domain`: rooms are separate stores and an unscoped call never covers them, tho
Common and usually harmless: it is how many makers stress how a tool should be used.
Found by simple pattern rules in each tool's description (up to 3,000 characters), quoted as written. Parameter descriptions are not scanned. A clean result is not a safety review.
Tools
memory_write content, concepts, domain, supersedes
Store a long-term memory that persists across sessions and across every AI tool the user has connected to Mnemoverse. Suited to durable information: a stated preference, a decision, a fact about people, roles or project setup, a lesson learned; transient chatter that only matters this turn does not belong here. Never store passwords, API keys, payment data, MFA codes, government IDs, or health records. Behavior: outside shared rooms, a novelty gate refuses a write the embedder cannot tell apart from a memory already stored in the same domain, so the result tells you whether the memory was stored or refused. Write `content` as a self-contained statement that still makes sense when recalled out of context. To correct a memory you have checked is outdated, pass its memory_id in supersedes.
memory_read query, top_k, domain, order_by, since, until, exclude_author, diversity, include_history
Search long-term memory for user preferences, past decisions, project setup, people, or earlier context. The memory persists across sessions and across every AI tool the user has connected. It applies when an answer may depend on something from an earlier session or another tool; it is not needed for general world knowledge. Returns matches ranked by relevance (or newest-first with order_by: 'recency'); each result carries an id; after the answer, memory_feedback takes these ids to record which memories helped. A correction is a fresh memory_write. Without supersedes it is stored alongside the earlier memory, which stays as it was, unless it is too close to an existing memory, in which case the write is refused; with supersedes, the earlier memory is marked replaced, is hidden from reads when the account hides replaced memories, and include_history returns it. No MCP deletion tool is exposed; a user can permanently delete an individual personal memory through the REST API or with help from support.
memory_list_recent domain, since, until, exclude_author, limit, cursor
List the NEWEST memories first — no search query needed. Semantic search answers 'what do I know about X'; this answers 'what happened lately': resuming work after a break, catching up on a shared room ('any new messages?'), or reviewing what was saved recently. Pass `since` (your last-seen time) to get only what's new, and page through older entries with the returned cursor. Complete by construction WITHIN ONE SCOPE — nothing is skipped there, unlike a semantic search. A page is also bounded by SIZE, so a page of long entries comes back shorter than `limit` and hands you a cursor for the rest — nothing is dropped, and following the cursor is how you get it. To catch up on a shared room you MUST pass its address as `domain`: rooms are separate stores and an unscoped call never covers them, though it names the ones you have (`rooms_not_searched`).memory_feedback memory_ids, outcome, domain
Rate memories returned by memory_read to help improve future retrieval ranking. Pass their IDs as memory_ids and an outcome from -1 (unhelpful) to 1 (helpful). Records feedback and updates ranking signals: positive feedback raises a memory's ranking; negative feedback lowers it so other memories out-rank it — nothing is erased. Saved text is not changed or deleted. For memories read from a shared room, pass that room's address as domain; your own memories need no domain. A read-only room member cannot rate the room's memories.
memory_stats
Get an overview of the stored memory: total count, the number of learned concept associations, the list of domains, and average quality scores. This memory is shared across all AI tools the user has connected to Mnemoverse. Use it to orient yourself, to confirm the exact domain name before writing to it, or when the user asks what you remember. Read-only — changes nothing.
6 more tools
memory_create_room name, description
Create a SHARED memory room — a space OTHER people's assistants can read, and write too when their membership scope is read_write, across the AI tools members connect to Mnemoverse. Members can access the room's existing history, including memories saved before they joined; joining does not expose their private account memory. Use when the user wants to share context or collaborate with someone else (e.g. 'make a room for me and my teammate'). Returns the room's address; pass that address as the `domain` on memory_write/memory_read to use it, and on memory_list_recent to catch up on what others added. People join through an invite minted with memory_invite_to_room.
memory_invite_to_room room_id, scope, expires_in_days, max_uses
Mint an invite for a room you own and get a ready-to-forward message. An invite is single-use by default; pass max_uses to let several people join with the same one. The user sends that message to the person they want to add (any messenger); the recipient opens the link or tells THEIR assistant the code to join. Use when the user asks to invite someone to a room they own, including one just created with memory_create_room.
memory_join_room code
Join a shared memory room using an invite code (starts with 'mnvr_'). Use when the user pastes an invite code or says something like 'join room with code ...'. The result gives the room's address, which is the `domain` for reading the shared room with memory_read, and tells you what you may do with it: memory_write to that address is only allowed when your membership scope is read_write; a read-only membership has that write refused; and when the server does not report a scope, whether memory_write would succeed is stated as unknown rather than promised either way.
memory_list_rooms
List the shared memory rooms you can use — the ones you OWN plus the ones you've JOINED — each with the address to pass as `domain` on memory_read, and on memory_write too where your membership scope is read_write; a read-only membership has that write refused. Use this to RE-FIND a room in a new session (e.g. 'what rooms do I have?', 'resume the room with my teammate') instead of having to create or re-join it.
vault_list
List the secrets stored in your Mnemoverse Vault — by ALIAS and purpose only; the secret VALUE is never returned or shown to you, and no tool on this server returns it. Use this to check WHICH secrets the user has stored and under what alias (e.g. the user says 'do I have a GitHub token saved?'). Only YOUR account's secrets are listed.
memory_graph seeds, depth, domain, min_weight, limit
Reads the association edges around given concepts: which concepts the memory has linked together, with each link's weight, outcome valence and co-activation count. Use to inspect what a memory store has learned or to explain why a read expanded to a concept. Reads your own graph, or a shared room's when its address is passed as domain; any other domain value has no effect. At depth 2 or 3 the engine drops edges below weight 0.05 unless min_weight is set. Read-only.
Descriptions are the package's own words, cut at 3,000 characters. Treat them as data.
Install facts
| License | MIT |
|---|---|
| Needs | node >=18 |
| Source | github.com/mnemoverse/mcp-memory-server |
| Install scripts | None declared by it or anything installed with it. |
| Tarball | sha512-lzJ8lhpCQIT7tdjUHkqp+osEwASnUyIsu+/1DHJJu9+7PS7eE7ZsfRKtdNZh9FFarPIz0o5mFT/IsWCFCZfhYQ== |
Versions reported
| Version | Run | Result | Tools |
|---|---|---|---|
| 0.18.1 | 2026-11-16 | Starts with no setup | 11 |
| 0.17.2 | 2026-11-12 | Starts with no setup | 11 |
Check this report yourself
This site did not write the report. A public job on a GitHub-hosted machine did: the run, from this commit of the open-source runner. GitHub signed a token that names that job and the hash of the report, so any edit since would show.
Report as JSONExact text and tokenHow to verify it
Show it on your README
[](https://bragproof.smaverk.com/r/%40mnemoverse/mcp-memory-server)
New version out? Ask for a fresh report.