E
Engram
Core Concepts

Knowledge Base

Turn Engram into a domain expert. Crawls external documentation sites, vectorizes the content into a local ChromaDB instance, and answers complex questions with citation-backed accuracy.

Recursive Crawler

Intelligently traverses documentation pages, respecting domain boundaries and separating code blocks from prose.

Local Vector Store

Uses ChromaDB to store high-dimensional embeddings on your disk. No data ever leaves your machine.

RAG Pipeline

Retrieval-Augmented Generation fetches the exact documentation snippet needed to ground the AI's answer.

Interface Options

Engram provides two ways to interact with your knowledge base: a visual dashboard for ease of use, and a CLI tool for automation.

Knowledge Base
https://flask.palletsprojects.com/
Ingest
How do I create a route in Flask?
AI: To create a route in Flask, use the @app.route() decorator.

# Example
@app.route("/")
def hello():
  return "Hello World"

Technical Setup

01Install Dependencies

The crawler requires specific Python libraries for scraping and vector storage. Ensure these are in your backend container.

text
beautifulsoup4
requests
chromadb

02Git Exclusion (Critical)

The vector database (`engram_db`) is heavy and contains generated data. You must exclude it from version control.

gitignore
# .gitignore
engram_db/
chroma.sqlite3