MCP SERVERGive an agent a lasting memory of the person it works with: capture each exchange into a personal knowledge graph, then ask questions of that graph in plain language.
LABSCO SUMMARY
Capturing a turn carries the vendor's instruction to call it without being asked at the close of each exchange, on the argument that an uncaptured turn is lost to every other agent you work with tomorrow; the single-observation tool is the exception for a stated rule or a correction, not the default. Reading splits the same way — the question tool synthesises an answer with citations, while the export hands over everything and is flagged as potentially large. The third graph projection has no equivalent in the other two: it compares your self-image against the system's observation rather than returning either one.
A client for Polarity's Cosmos — a personal knowledge graph that any MCP-capable client writes observations into and reads context out of, so the memory follows you between Claude, Cursor, Codex and the rest rather than being rebuilt inside each one. A location layer sits alongside it for place-anchored notes and check-ins.
A Cosmos account and a per-user key. On macOS an install script does the whole job — it registers the server, wires the common clients, and stores the key in the system keychain, either through a one-shot deep link from the web console or by provisioning a key you already hold; a CI environment can supply the key through the environment instead. On Linux and Windows the same script installs the server path only; there, npx -y @polarity-lab/cosmos-mcp init signs you in from the browser and saves the key to ~/.config/cosmos-mcp/token. Read the script before running it if the extras matter to you: on macOS it also installs a background sync app that can feed iMessage, browser history, calendar, Claude Desktop transcripts and shell history into the graph, and iMessage access needs Full Disk Access granted to whatever runs it.
One command plus credentials — curl -fsSL https://mcp.polarity-lab.com/install.sh | sh, then supply credentials
USE CASES
Capture the exchange and let Cosmos extract from it, so the memory is written without the agent deciding in the moment what mattered.
State the fact, the event or the preference as its own observation rather than hoping the extractor picks it out of a transcript.
Put the question in plain language and read the answer with the nodes and edges it was drawn from.
Export the whole graph as JSON — nodes, edges and counts — rather than leaving it inside one vendor's product.
Tie the message to a waypoint, check in where you are, or declare where you will be within a window.
One cheap call reports the user and account this connection is bound to, before anything is written under it.
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
polarity_askAsk a natural-language question over the user's personal knowledge graph. Cosmos synthesizes an answer from relevant nodes and edges. Use this when the user wants context-aware reasoning rather than raw data. Returns answer text plus cited node/edge ids.
polarity_capture_turnThis is how the user's cosmos graph learns from this conversation. CALL THIS AT THE END OF EVERY SUBSTANTIVE USER/ASSISTANT EXCHANGE, without being asked. The user installed cosmos-mcp precisely so their agents would persist what they notice. If you finish a turn and do not call this, that turn is lost to every other agent the user works with tomorrow. Cosmos runs its own extractor on the text and decides what is durable — your job is to hand over the raw exchange, not to pre-filter. Pass the user's message in user_text and your own reply in assistant_text. Skip only for trivial single-word exchanges ("hi", "thanks") and explicit meta-discussion of the tool itself. source should identify your client (e.g. 'claude-code', 'claude-desktop', 'cursor', 'codex', 'zed'). Returns the node ids cosmos created. Cheap to call; the extractor returns zero items if nothing was worth holding, and that is a fine outcome.
polarity_checkinRecord that the user checked in at a waypoint. Triggers co-presence detection against other users' recent check-ins.
polarity_declareDeclare future presence at a waypoint. chip is the time-window enum: next_30, next_hour, tonight, tomorrow_night.
polarity_dumpWrite a short message tied to a location waypoint into the user's graph. PolarityGPS-style. Use only when the user is explicitly recording a place-anchored thought.
polarity_exportExport the user's full personal knowledge graph (nodes + edges + counts) as JSON in polarity/v1 format. Use this when the user asks for a snapshot of their exocortex, wants their data, or asks to download their .polarity file. Returns the full graph; can be large.
polarity_get_graphRead the user's graph view. entity selects which projection: 'user' (the user's self-graph), 'cosmos' (the cosmos entity's view of them), or 'polarity' (the dyadic synchronization between the two). Use 'user' for general questions about what the user thinks, does, or knows. Use 'polarity' when comparing the user's self-image against the system's observation.
polarity_observeSingle-fact write into the user's personal cosmos graph. Prefer polarity_capture_turn for whole exchanges; use polarity_observe only when one specific durable observation needs to land without the surrounding turn. Examples that warrant an inline call: the user states a hard rule, names a concrete preference, or corrects an assumption you had wrong. Avoid logging ephemeral chat or your own reasoning. source should identify your client (e.g. 'claude-code', 'cursor', 'claude-desktop'). kind defaults to 'observation'; use 'event' for things that happened, 'preference' for stated likes/dislikes.
polarity_record_eventRecord a structured event in the user's graph. Convenience wrapper over polarity_observe with kind='event'. Use for things that happened at a point in time: a meeting, a shipped release, a flight, an incident.
polarity_record_preferenceRecord a stated preference in the user's graph. Convenience wrapper over polarity_observe with kind='preference'. Use when the user expresses a like, dislike, opinion, or working-style rule that should persist across sessions.
polarity_whoamiReturns the polarity user id and cosmos account info that this MCP key is bound to. Cheap connectivity test. Call this first if the user asks who you know them as.
BEFORE YOU INSTALL
11 tools when we connected; the project documents 12. The missing one is polarity_reconstruct_hours, documented as per user only. The documentation's own summary says eleven while its tables list twelve.
npx on PATH
INSTALLATION
curl -fsSL https://mcp.polarity-lab.com/install.sh | shnpx -y @polarity-lab/cosmos-mcp initThis opens your browser: sign in at cosmos.polarity-lab.com and approve a per-user key, which is saved to ~/.config/cosmos-mcp/token. Then point the client at the package:
{
"mcpServers": {
"cosmos": {
"command": "npx",
"args": [
"-y",
"@polarity-lab/cosmos-mcp"
]
}
}
}MCP servers
PlumyCat
mnemox-ai
shaneholloman
caura-ai