Context & Memory Overview

Every time you send a message in Grepp, the agent automatically builds a complete picture of your current situation before responding. You don't need to repeat yourself or re-explain your data — Grepp handles this context behind the scenes.

What's included in your context

Each conversation turn gives the agent access to:

• Your conversation history — everything discussed so far in the session, so you can ask follow-up questions naturally
• Your database schema — the tables, columns, and types from any connected databases, so the agent can write accurate queries without you describing your data model
• Your dashboard state — the charts and components currently on your dashboard, so the agent knows what already exists and can reference or update them
• Your installed integrations — which data sources are connected and available to query
• Work done this session — any charts or dashboards created or modified during the conversation, so the agent can reference them by name or ID without re-querying

Why this matters

Because the agent always has this full picture, you can work naturally:

• Ask "make that chart a bar chart instead" without specifying which chart
• Say "add a revenue column to that query" without re-explaining the query
• Request "save this to my dashboard" without describing what was just created

See also:

• Conversation History — how chat history is preserved across turns
• Schema Awareness — how Grepp understands your database structure
• Multi-Turn Context — how the agent handles follow-up questions and refinements
• Recall — how the agent remembers previous conversations and past work