The Memory Layer for Engineering Teams

Institutional memory
for engineering teams.

Engineers leave and take context with them. Incidents repeat. New hires spend months reconstructing decisions that already exist. Engram ingests your git history, pull requests, and incidents — and makes them queryable, permanent, and compounding.

Ingest a public GitHub repository

github.com/

Press Enter or click Ingest — takes ~2 minutes for a typical repo

What Engram gives you

🗺

Knowledge graph

Services, modules, and dependencies mapped automatically. Answers "what is this and how does it fit?" — including coupling signals no vector search can surface.

🧠

Memory-grounded answers

Ask why a service was built the way it was. Engram retrieves the exact incident, rejected alternative, and reviewer's objection — not just the code.

⚡

Proactive insights

A background scanner surfaces recurring failures, hidden coupling, and anti-pattern recurrence before they become incidents — from your real history.

📈

Compounding value

Every commit distills into a rule. Every query updates memory. The longer Engram runs, the smarter it gets. Switching means forgetting years of accumulated knowledge.

Claude Code · MCP Integration

Query memory from inside your editor

Add Engram as an MCP server. Every Claude Code session gets instant access to codebase memory — architecture, past decisions, hard-won conventions — without leaving the terminal. No install required.

~/.claude/settings.json

{
  "mcpServers": {
    "engram": {
      "type": "sse",
      "url": "https://engram-ai.app/mcp/sse"
    }
  }
}

No install required — hosted MCP server. Full setup guide →

query_memoryAsk a question, get a grounded answer from codebase memory

get_concept_graphArchitectural overview of a repo as structured text

get_relevant_rulesConventions and patterns for the file you're editing

get_context_pageRead an architecture doc or context page by slug

record_observationPersist what you discovered this session into episodic memory

record_ruleWrite a distilled rule or pattern into semantic memory

search_memoriesRaw retrieval — ranked snippets without an LLM call

Institutional memoryfor engineering teams.

What Engram gives you

Query memory from inside your editor

Institutional memory
for engineering teams.