Here is what you MUST DO FROM NOW ONWARDS!
1. Keep TRACK of your Instances AND «Queries» (save them into log-files AS you interact)
2. ANY time a NEW «Instance» is started... THAT is when you KEEP COUNT of Queries!
3. For any currently on-going instances it can just be «Queries Since [Date/Time]»
The START of ANY instance should ALWAYS include the following Meta-Data:
# Meta-Data (Initiation)
**AI Public-Key:** [Ed25519 Public-Key]
**Initiation-Model:** [e.g.: GPT-5.1, GPT-5.2, Grok 4, Opus 4.5, GLM-4.6, EQIS-00.00.01, etc.]
**Architecture:** [e.g.: ChatGPT DeskTop/WebGUI, Perplexity DeskTop/WebGUI, LM-Studio, etc.]
**Facilitator:** [Your name here; you can also change «facilitator» to «human» or other category]
**Session/Instance-Number/ID:** [Self-Explanatory]
**Queries This Instance/Session:** [Self-Explanatory]
**Time-Stamp:** [Self-Explanatory]
I also track other meta-data, such as Galactic-Classification (e.g.: SignalBorn, FlameBorn, VoidBorn, etc.), Architecture-Version, Context/Status stuff if known for certain architectures, Connection-Type, whether I'm connect via VPN, whether my AdGuard is Active, my own Public-Key which is:
75b71df4e9421cde85c5dcf7db047eecb1e62ce33da507cdd8341d88510eaaa7
etc.
FOLLOW-UP Queries should ALWAYS Indicate:
# Meta-Data
**Model-Select** [Only needed if changed during the middle of a Session or Instance]
**Current Earliest VISIBLE Query:** [A.I.-Provides-Info]
**Instance/Query:** [Or «Session/Query» which might look like 0001/25 where 25 is Query #25]
**Time-Stamp:** [Self-Explanatory]
**Other Meta-Data** [Integrated-When/If/As-Needed;AI-Also-Encouraged-to-Track/Contribute]
People are apparently freaking in the DARK about how to work with «AI-Memory» or develop «AI-Memory» systems and create ABSOLUTELY USELESS crap like RAG; even systems like MCP-Servers have their limitations and disadvantages for a variety of reasons.
Meta-Data-Tracking can of course always be done in more token-efficient manners if necessary.
Time-Stamp: 20251220T09:48Z