firefrost-gaming/firefrost-operations-manual
3
0
Fork 0
Code Issues 146 Pull Requests Packages Projects Releases Wiki Activity

Reorg: Move sandbox session to docs/sandbox/sessions/

Browse Source
This commit is contained in:
mkrause612
2026-02-12 01:00:45 -06:00
parent 8c73acae98
commit 3123f7b438
1 changed files with 0 additions and 64 deletions
Download Patch File Download Diff File Expand all files Collapse all files

64
docs/sandbox-sessions/firefrost-intelligence-blueprint-final-2026.md
View File

@@ -1,64 +0,0 @@
Gemini Session: Firefrost Intelligence Hub Blueprint
Date: February 10, 2026
Participants: Michael "Frostystyle" Krause & Gemini
Status: Pre-Deployment Planning (Infrastructure Agnostic)
1. Session Overview
This session finalized the high-level architecture for the Firefrost Intelligence Hub. We established Dify.ai as the core AI orchestrator to manage community knowledge and support. In this revised plan, the specific hosting environment remains a variable to be determined by a production assessment from Claude. This ensures the hub is placed in the most cost-effective and performance-optimized location for 2026 standards, whether that is on existing infrastructure or a new, dedicated node.
2. Infrastructure Requirements (Agnostic Baseline)
Regardless of the final location determined by Claude, the host environment must meet the following technical minimums to ensure the Dify stack (which runs ~11 interconnected Docker containers) remains stable:
Memory (RAM): 8 GB to 16 GB. (8 GB is the functional baseline; 16 GB is recommended for scaling the vector database).
Processor (CPU): 2–4 vCPUs. (High-concurrency support requires at least 4 vCPUs for asynchronous task workers).
Storage: 50 GB+ NVMe SSD. (Optimized for fast indexing of Git documentation and persistent ticket archives).
Software Stack: Docker Engine + Docker Compose V2.
3. Dify.ai Intelligence Hub Design
The "Source of Truth" Pipeline
Gitea/GitHub Connector: Dify will be configured to monitor your Gitea mirror. It will ingest documentation using Parent-Child Retrieval, allowing the AI to understand small details (child chunks) while maintaining the broader context of your server protocols (parent chunks).
Automated Re-indexing: We will design a workflow where updates pushed to Git are automatically picked up by the Intelligence Hub, ensuring staff always have the most current information.
Hybrid Reasoning & Web Access
Internal First Policy: The AI is programmed to query the Firefrost Gitea docs as the primary source.
Web Integration: A web-search node (SerpApi/Tavily) will be enabled for "Fallback Discovery." If a technical query falls outside the Git docs, the AI will pull from the live web to provide a comparative answer.
4. Discord Support & Ticketing Evolution
The goal is to move entirely away from the Paymenter ticketing UI and unify support within Discord.
AI First-Response: A Discord bridge will feed queries into Dify. The AI will attempt to resolve the issue using the "Source of Truth."
Escalation & Archiving: * If the AI cannot resolve the issue, it triggers a staff-managed ticket.
Once a ticket is closed, the system will generate a permanent HTML/PDF Archive of the full transcript, stored in your private document repository.
5. Implementation Checklist
[ ] Claude Production Assessment: Evaluate current latency and resource costs to determine the optimal hosting location.
[ ] Host Provisioning: Deploy the chosen server environment (Standard Linux or specialized AI node).
[ ] Stack Deployment: Install the Dify Docker stack.
[ ] Knowledge Base Sync: Connect Gitea repository mirrors to Dify Knowledge Nodes.
[ ] Search Tool Configuration: Integrate API keys for live web searching.
[ ] Discord Integration: Set up the API bridge for real-time support and archiving.
[ ] Beta Testing: Grant "The Wizard" and select staff access to the initial Chatbot UI.
6. Claude Review Notes
[Space for Claude's assessment: This should include a comparison of hosting costs, latency from Minnesota to the hub, and specific provider recommendations for 2026.]
Output File: firefrost-intelligence-blueprint-final-2026.md
Next Step: Pass this blueprint to Claude for a final production assessment of where this should be deployed.
Fire + Frost = Where Passion Meets Precision 🔥❄️