Guillaume, 2026-10-10, session miadi-episodic-memory-schema-and-mw-261010. His words, trimmed of paths that only name files.
First message:
"packages/episodic-memory-schema + medicine-wheel-upstream probably. app/chronicle. packages/ep120-factory-interface-analysis/app was created and I will need some help in a relationship to potential relation into our chronicle.
First: one for prototype like packages/ep120-factory-interface-analysis/app published at https://miadi-factory-interface-analysis.vercel.app/ for miadi-chronicle://120 so that the published URL of the prototype would be accessible for most of our users and the editing URL https://v0.app/jgwills-projects/chat/miadi-factory-interface-analysis-g6dimPe7PBM would be accessible to admin and other relevant roles.
Second: the deepdiver has added to miadi-chronicle:550 some of his work and maybe the 'notebooklm URL' he uses is already added in either the episodic-memory-schema or the wheel; if that is the case nothing to do, otherwise todo.
Prompt/traces/sessions/datasets/score config in miadi-chronicle. Third is about supporting the connection to langfuse so that we can associate prompts and traces from langfuse to episode (as probably implied in the prototype itself, that we'd implement fast). Eventually, we would use 3 different types of tracing per domain; we'd start with feeding ourselves for ceremony, environment variables CEREMONY_LANGFUSE_SECRET_KEY, CEREMONY_LANGFUSE_PUBLIC_KEY, CEREMONY_LANGFUSE_BASE_URL, and in the future STORY_LANGFUSE_* for story writing itself (ncp-wampum-bridge, ncp-story-studio, app/ncp) and ENG_LANGFUSE_* for engineering level stuff (hooks-core, hooks-gateway, hooks-interpreter, or the miadi-session-observability plugin would trace agent sessions in that tracing, not the ceremony one). Engineering in the sense that we use that one for getting better and have traceability which we don't have now. Feels like we shall think about doing that now.
Fourth is hard to comprehend but relates runtime/tide-runtime and the app/tide cockpit to an episode. Will relate to all plugins and miadi-orchestration-kit/teams/README.md, pretty sure.
Five is more complex. The other thing I'd like to be capable to associate (like we are already capable to associate a foundation, issues/inquiry, miadi-review, talking-circle, ceremony, stateloom-diagrams, and I might forget some) would be created/updated RISE Framework specs, but the RISE framework would need touch-up as we evolved so much our ways of working. I know that it does not save like I want (ex. rispecs/<subfolder feature/capability related>/.[spec|story].md). I don't remember which of the miadi-chronicle envisioned including the user stories. We upgraded smcraft (stateloom-diagrams, that we already relate to episode) to support Entity-Relationship-Diagram and Sequence-Diagram, and my strong feeling is that the whole RISE framework needs something that would complete the cycle that is even implied in the prototype that motivated this input. I want prototypes we create to be part of episodes.
It might imply upstream work on jgwill/medicine-wheel and the RISE framework in jgwill/llms-txt. Use carefully subagents to do a careful job. Sit with this and come back with something useful."
Second message, after the proposal was published and after I said I had not used Jev:
"In the job that you're doing, it's really important, and really new for me, that new class of model. I really feel that by combining your intelligence with that other class of model, you'll be able to revise your proposal with scoring. In the future I'm expecting agents such as yourself to be able to take decisions on a level of confidence on what will be executed and implemented. Your proposal would be: here is what I would do that I'm certain about, and another section where you're not certain about. You're developing the system as well. I don't know exactly where you could store what we're talking about, but I want you to self-evolve and be able to use that, and other new instances in the repositories and inside the factory will be able to reuse that intelligence. I started in other sessions to achieve that a little bit. Other instances are working on the writing platform and the story studio; be careful not to step on their toes if you're saving files."
Third message:
"There's probably an episode that talks about Jev, so that could be where you relate this: maybe just a GitHub issue that you weave into that episode, and inside that issue you document this self-evolution and what we are doing here, because my first prompt could not contain the whole narrative: the proposal, and me asking you to use the decision engine and self-evolve to be capable of helping the whole platform, deciding what is obvious. My feeling is that by using that engine you'll be able to take decisions and come back to me just when this is actually needed."