Allows routing memory storage to your own Supabase PostgreSQL (pgvector) database by supplying a Session Pooler URL, so ingested events and memories are kept entirely in your own infrastructure instead of Smriti's cloud (BYODB mode).
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
The Problem: Traditional "agent memory" relies on fuzzy vector databases that suffer from temporal drift. When tracking deals, deployments, contracts, or multi-step workflows, you need a deterministic event ledger — not a chatbot guessing from old embeddings.
The Solution: Smriti is a temporal event memory layer — a bi-temporal hippocampus for AI agents that tracks state changes perfectly across systems and sessions.
Feature | Smriti (Event Memory) | Traditional RAG (Chat Memory) |
Structure | SVO Tuples: Tracks concrete actions ( | Vector Sludge: Dumps unstructured text into a vector DB |
Time | Bi-Temporal: Tracks both when it happened and when it was recorded | Timeless: Cannot distinguish old facts from current state |
Updates | Supersession: Safely overrides state without deleting history | Overwrite: Hard deletes or creates confusing duplicates |
Related MCP server: brain
Step 1 — Get a free API key
curl -X POST "https://smriti-kaal.vercel.app/billing/keys?tier=explorer"Step 2 — Store a memory
curl -X POST https://smriti-kaal.vercel.app/ingest \
-H "X-API-Key: chrn_your_key" \
-H "Content-Type: application/json" \
-d '{"source_id": "demo", "events": [{"text": "Alice joined as Lead Engineer on July 15"}]}'Step 3 — Recall it
curl -X POST https://smriti-kaal.vercel.app/query \
-H "X-API-Key: chrn_your_key" \
-H "Content-Type: application/json" \
-d '{"query": "Who joined the team recently?"}'Response:
{
"results": [
{
"subject": "Alice",
"verb": "joined",
"object": "the team as Lead Engineer",
"timestamp": "2026-07-15T00:00:00Z",
"confidence_score": 0.94
}
]
}Smriti ships with a built-in Model Context Protocol (MCP) server. It runs as a lightweight local process — no backend to host, no infrastructure to manage.
pip install smriti-kaal-mcp
# or zero-install via uvx (no pip needed):
uvx smriti-kaal-mcpAdd to claude_desktop_config.json:
{
"mcpServers": {
"smriti": {
"command": "uvx",
"args": ["smriti-kaal-mcp"],
"env": {
"SMRITI_API_KEY": "chrn_your_key_here",
"SMRITI_SOURCE_ID": "claude-desktop"
}
}
}
}Add to .cursor/mcp.json in your project root:
{
"mcpServers": {
"smriti": {
"command": "uvx",
"args": ["smriti-kaal-mcp"],
"env": {
"SMRITI_API_KEY": "chrn_your_key_here",
"SMRITI_SOURCE_ID": "cursor-project"
}
}
}
}Tool | What It Does |
| Store text — auto-extracts causal SVO events |
| Hybrid search (semantic + temporal) across all memories |
| Generate a chronological timeline of events |
| Supersede outdated memories cleanly |
| View service health and DB status |
Keep your memory data in your own infrastructure using Supabase (PostgreSQL + pgvector). Just pass your Session Pooler URL as a header — no code changes required:
curl -X POST https://smriti-kaal.vercel.app/ingest \
-H "X-API-Key: chrn_your_key" \
-H "X-Supabase-Url: postgresql://postgres.xxxx:YOUR_PASSWORD@aws-0-REGION.pooler.supabase.com:5432/postgres" \
-H "Content-Type: application/json" \
-d '{"source_id": "my-app", "events": [{"text": "Alice joined the team"}]}'👉 Full 3-step BYODB setup guide →
When you send text to Smriti, it passes through four discrete layers:
Ingest Gate — Accepts raw text from chat logs, emails, or code
Decomposition — LLM extracts structured Subject → Verb → Object tuples
Temporal Binding — Assigns exact temporal boundaries to track state changes over time
Dual-Store Indexing — Saves to PostgreSQL (exact chronological queries) + pgvector (semantic search)
Base URL: https://smriti-kaal.vercel.app
Auth: X-API-Key: chrn_your_key
Endpoint | Method | Purpose |
| POST | Generate a new API key |
| POST | Convert raw text into structured memories |
| POST | Natural language memory retrieval |
| POST | Chat with the memory-aware LangGraph agent |
| GET | System uptime and DB status |
const headers = { "X-API-Key": "chrn_...", "Content-Type": "application/json" };
await fetch("https://smriti-kaal.vercel.app/ingest", {
method: "POST", headers,
body: JSON.stringify({ source_id: "app", events: [{ text: "Upgraded to Pro" }] })
});
const res = await fetch("https://smriti-kaal.vercel.app/query", {
method: "POST", headers,
body: JSON.stringify({ query: "Who upgraded?" })
});
console.log(await res.json());import httpx
API = "https://smriti-kaal.vercel.app"
HEADERS = {"X-API-Key": "chrn_..."}
httpx.post(f"{API}/ingest", headers=HEADERS, json={
"source_id": "my-app",
"events": [{"text": "User completed onboarding on July 15"}]
})
result = httpx.post(f"{API}/query", headers=HEADERS, json={"query": "What did the user do?"})
print(result.json()["results"])Variable | Required | Description |
| ✅ | Your API key ( |
| No | Namespace for your memories (e.g. |
| No | Session Pooler URL to route memory to your own Supabase DB |
| No | Memory scope ( |
| No | Override the API base URL (advanced) |
Tier | Price | Limits |
Explorer | Free | 10,000 events/mo · 3 connected tools |
Builder | $49/mo | 500,000 events/mo · 25 tools |
Scale | $249/mo | 5,000,000 events/mo · unlimited tools |
Manage keys and view memory timelines at smriti-kaal.vercel.app
All API requests require a bearer token (X-API-Key)
API keys are scoped per source_id — one application cannot read another's memories
BYODB mode: your data never touches Smriti's cloud storage
MCP server runs locally on your machine; only outbound HTTPS calls are made
MIT © 2026 Smriti / Chronos Labs
This server cannot be deployed
Truth-validated, SHA-512-sealed AI memory for Claude & Cursor. Free tier, OAuth, 38 tools.
Private memory for AI agents: save and recall facts across Claude Code, Codex, Cursor and ChatGPT.
Long-term memory for Claude, ChatGPT, Cursor: search, save and organize notes, links, docs, videos.
Persistent memory for Claude Code and Cursor. Stop re-explaining your project every session.
