🏗️ Architecture: How should Vitto's chat gain self-awareness about its own Mi...

Date: 2026-02-11
Time: 19:33:16 UTC
Session: gspace1-105.25
Agent: backend
Confidence: 95%
Status: ✅ Documented

📋 Decision Context

Question:

How should Vitto's chat gain self-awareness about its own Mind state?

Task: Session 105.5: Vitto Mind-aware chat with self-knowledge injection

🎯 Decision Made

Chosen:

Mind context injection on first message: VittoChatService calls GET /engine/mind/summary, formats it as a compact text block, and injects it into the system prompt. Cached per conversation. This gives Vitto self-knowledge without requiring backend changes to the chat endpoint.

Options Considered:

• No alternatives documented

💭 Reasoning

The Mind summary endpoint already exists (created earlier in this session). By injecting it into the system prompt, Vitto can reference its own stats, lessons, and codebase knowledge naturally in conversation. The ~200 extra tokens are worth the intelligence gain. Caching per conversation prevents redundant API calls.

📁 Files Involved

• libs/libs-ai/gograb-ai-lib/src/lib/services/vitto-chat.service.ts

🔗 Related

• Decision ID: dev_1770838396511_knp9u5l
• Session: 105
• Workspace: gspace1

Auto-generated by MCP Smart Consolidation