Skip to content
Engram

Your First Memory

Once Engram is installed and connected to your AI client, you can start building your knowledge graph immediately. Your AI agent will use Engram’s MCP tools to remember, recall, and connect information across sessions.

Store a Memory

Section titled “Store a Memory”

Ask your AI to remember something:

"Remember that our API uses JWT tokens with 15-minute expiry and refresh tokens stored in httpOnly cookies."

Behind the scenes, Engram calls the remember tool:

{
  "content": "API uses JWT tokens with 15-minute expiry and refresh tokens stored in httpOnly cookies",
  "project": "my-project",
  "topic": "architecture"
}

The memory is stored with a semantic embedding, enabling recall by meaning rather than exact keywords.

Recall a Memory

Section titled “Recall a Memory”

Later, in a new session, ask about authentication:

"How does our auth system work?"

Engram finds the relevant memory via semantic search — even though you never said “auth” in the original memory.

Connect Memories

Section titled “Connect Memories”

As you store more memories, connect related ones to build a knowledge graph:

{
  "from": 1,
  "to": 2,
  "relation": "relates_to"
}

Supported relation types: relates_to, supersedes, contradicts, part_of, derived_from, summarized_by.

Next Steps

Section titled “Next Steps”