Dejaview Mcp
About
DejaView is a persistent knowledge graph memory layer for AI agents. It gives every MCP-compatible AI host shared, queryable memory of people, projects, decisions, and relationships — stored in a graph that grows smarter over time.
Details
- Author
- JakeC77
- Downloads
- 187
- Categories
- Knowledge Base
Jump to
- Knowledge graph with typed relationships between any entities
- Natural language Q&A with answers cited from stored facts
- Public sharing of entity subgraphs as interactive visual graphs
- Works with Claude Desktop, Cursor, Windsurf, and any MCP host
- Tools for storing, recalling, searching, and forgetting facts
- Timeline and graph statistics tools for memory overview
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
Dejaview McpCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Install with pip install dejaview-mcp, then add a dejaview entry to your MCP server configuration with the DEJAVIEW_API_KEY environment variable. The server exposes 12 tools for storing, recalling, searching, and managing graph memories; call agent_context at the start of every conversation to load a full memory summary.
agent_context
Get a full context summary of your knowledge graph. Returns entity counts, most-connected entities, and recent activity. Call this at the start of every conversation to bootstrap your memory.
remember
Store a fact as subject -> predicate -> object in the knowledge graph. Use for any relationship, preference, decision, or event worth keeping across sessions. Examples: remember("Alice", "works_at", "Orbit Labs") remember("Project Atlas", "has_status", "in progress", "As of Q1 2026")
remember_many
Store multiple facts at once. More efficient than looping remember(). Each fact needs: subject, predicate, object. Optional: context.
recall
Get everything known about an entity — all relationships in and out. Use for full context on a person, project, org, concept, or any entity.
search
Search for entities in the graph by name (partial match). Use this to discover what's in the graph before calling recall(). Returns names, types, and connection counts.
ask
Ask a natural language question about your knowledge graph. Returns a synthesized answer backed by cited graph facts — no hallucination. Every claim traces to a real stored fact with a timestamp.
timeline
Get recent facts in reverse chronological order. Use at session start alongside agent_context() to see what was recently remembered.
graph_stats
Get a high-level summary of the knowledge graph. Returns entity count, entity types, relationship count, and relationship types.
share
Create a public shareable link for any entity's subgraph. Returns a URL anyone can open — no account needed. Shows an interactive visual graph of the entity and its connections.
forget
Remove a specific fact from the knowledge graph. Deletes the relationship between subject and object but leaves both entities intact. Use to correct wrong or outdated facts.
forget_entity
Remove an entity AND all its relationships from the knowledge graph. Use with care — this permanently wipes the node and every edge connected to it. Good for removing entirely wrong or duplicate entities.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"dejaview mcp": {
"dejaview": {
"type": "streamable-http",
"url": "https://api.dejaview.io/mcp",
"headers": {
"Authorization": "Bearer dv_your_key_here"
}
}
}
}
}McpServers
{
"dejaview": {
"type": "streamable-http",
"url": "https://api.dejaview.io/mcp",
"headers": {
"Authorization": "Bearer dv_your_key_here"
}
}
}DejaView is a persistent knowledge graph memory layer for AI agents. It gives
every MCP-compatible AI host shared, queryable memory of people, projects,
decisions, and relationships — stored in a graph that grows smarter over time.
Get your API key at https://dejaview.io
Setup
Install:
pip install dejaview-mcpAdd to your Claude Desktop, Cursor, or Windsurf config:
```json
{
"mcpServers": {
"dejaview": {
"command": "dejaview-mcp",
"env": {
"DEJAVIEW_API_KEY": "dv_your_key_here"
}
}
}
}
12 MCP Tools
Tool Description
agent_context Load a full memory summary — top entities, recent activity, graph stats. Call at the start of every conversation.
remember Store a fact as subject → predicate → object. Use for decisions, relationships, preferences, events.
remember_many Store multiple facts at once. More efficient than looping remember().
recall Get everything known about an entity — all relationships in and out.
search Find entities by name (partial match). Use before recall() when unsure of exact name.
ask Ask a natural language question about your graph. Returns a cited answer backed by real stored facts.
timeline See recently stored facts in reverse chronological order.
graph_stats High-level summary: entity count, types, relationship count.
share Generate a public interactive graph link for any entity and its connections.
forget Remove a specific fact without deleting the entities.
forget_entity Remove an entity and all its relationships.
Key Features
Knowledge Graph — typed relationships between any entities: people, projects, orgs, tools, decisions
Natural Language Q&A — ask questions in plain English, get answers with graph citations
Public Sharing — share any entity's subgraph as an interactive visual graph, no account needed
Works everywhere — Claude Desktop, Cursor, Windsurf, or any MCP-compatible host
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
