Home / Ops / More in this area

Session State + Terminal Startup Guide (v1)

Updated Jul 08, 2026 · Affirmology_SessionState_and_Startup_v1.md

Session State + Terminal Startup Guide (v1)

Saved 2026-07-08 for Jeff, at a session reboot. What is done and durable, and what (if anything) you need to start in a terminal. No em dashes.

WHAT'S READY / DURABLE (nothing lost on reboot, all cloud-hosted)

Audios (permanent cloud links, also emailed): - Sol, promise-led (newest, all her feedback applied): Recognition 10:38, Design 11:22. - A / B / C Mirror for you and Sol (6 more). - Mission video compressed 590MB -> 85MB, live on R2 at experience/mission-video.mp4.

Foundations (cloud-stored): - DEEP six-lens Understandings for you (12,162 words) and Sol (10,756 words), stored SEPARATELY in the knowledge base, so your profiles keep the standard Understanding for testing. - ~13+ standard Understandings across the list, stored on their profiles. - Craft knowledge base wired into script generation, deployed to the cloud.

Specs / decisions written: - Cloud audio-creation spec (the real fix): Affirmology_CloudAudioCreation_Spec_v1.md. Lets you create audios/variations in the cloud with a selectable SOURCE (profile / understanding / deep understanding), handed to the code chat. - Mirror method + concepts + feedback: Affirmology_Mirror_v2_Methodology_v1.md, Affirmology_MirrorConcepts_Five_v1.md.

WHAT TO START IN A TERMINAL

The LIVE system: NOTHING. It is all in the cloud, always on.

Verified 2026-07-08: - Cloud Studio API (the agents, demo bridge, profiles, Understanding storage): https://affirmology-studio-api.onrender.com/api/health -> status ok. Runs on Render, always on. You start nothing. - Demo site https://demo.affirmology.ai -> HTTP 200. - The iOS apps (Atlas, Maya) talk to that cloud. Nothing local needed. So after a reboot, to have the live product working: open nothing, start nothing. The cloud is up.

The OLD local server (port 8765): ignore or kill it.

A legacy FastAPI server may still be listening on :8765 from the old laptop-tunnel architecture. It is RETIRED. The Tailscale funnel to it is dead; nothing routes to it anymore. You do not need to start it. To clear it if it lingers after reboot:

lsof -ti:8765 | xargs kill    # optional cleanup; nothing depends on it

The ONLY local dependency: the render env I use (a stopgap, not a service).

Generating Understandings and rendering Mirror audios has been running via Python scripts on THIS Mac. That is the thing we are eliminating with the cloud audio-creation endpoint. It is a bootstrap I run inside a session, not a service you manage. If a session needs to render locally before the cloud endpoint exists, the one-time setup is:

# ffmpeg (once)
brew install ffmpeg

# engine python env (persistent, survives reboot)
python3 -m venv ~/.affirmology-venv
~/.affirmology-venv/bin/pip install requests python-dotenv pydantic pyswisseph geopy timezonefinder boto3 anthropic

# run engine scripts with the studio-vendored engine on the path + the local corpus:
PYTHONPATH=/Users/jeffreyparker/CLAUDE/AFFIRMOLOGY/affirmology-studio/affirmology-agent/src \
AFFIRMOLOGY_CORPUS_DB=/Users/jeffreyparker/CLAUDE/AFFIRMOLOGY/corpus/corpus.db \
~/.affirmology-venv/bin/python <script.py>

Keys are already in affirmology-agent/.env and affirmology-studio/.env (Anthropic, Fish, ElevenLabs, R2, Resend). The corpus.db is at corpus/corpus.db (pull the current one from R2 if stale).

WHAT'S NEXT (for when you're back)

1. Build YOUR two (Recognition + Design) from your deep file.
2. Finish the list Understandings (the batch dropped partway on wifi).
3. BUILD THE CLOUD ENDPOINT (POST /api/create-audio, source-selectable) per the spec, so none of this touches a laptop again. This is the priority so a terminal session is never required for creation.

Bottom line: the product runs in the cloud with nothing to start. The only reason a terminal has been involved is the stopgap local generation, which the cloud endpoint removes.