Session Handoff — YouTube Content Engine: scoped, built PRD + Fable prompt, merged two tabs

Where it started

Session opened with /prime. Sondra is in Sandy Lee's Skool class (sandyleeai), watched Sandy's "AI Content Team" build video, and asked how to set up that engine for her own YouTube. It evolved: pull the class → realize she owns ~80% of the stack → she rejected a Google-Sheet dashboard → showed Sandy's product (sleestudio.com) → decided she wants her OWN version, for her, multi-channel + faceless from the start, delivered as a PRD + a Fable build prompt via the cloneify skill. She then pasted a handoff from a parallel tab and asked to combine everything (no duplicate effort), and finally questioned whether the roast verdict was based on stale framing.

Decisions locked + what shipped

• Sandy's class pulled → c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\Sandy Lee AI Content Team\classroom\ (62 lessons, 6 videos, 7 .skill files) + extracted-skills\. Command used: python scripts/skool_classroom_download.py --community sandyleeai --course-id fcecf44f --out "PROJECTS/Sandy Lee AI Content Team/classroom".
• Phase-1 "Content Team" prototype shipped in PROJECTS\Sandy Lee AI Content Team\: PRD.md, CLAUDE.md control panel, brand-voice.md (DRAFT — needs Sondra's edit), content-team\RUN-TREND-SCOUT.md, and a real digest digests\2026-07-06-digest.md. Trend Scout proven live via vidIQ MCP against her channel (UCOVr7yl6GJwrtdoWea3lrBw, 1,070 subs).
• The real project = YouTube Content Engine (her own Slee Studio, private, local, multi-channel face|faceless). Full PRD + cloneify run + Fable prompt written to PROJECTS\YouTube Content Engine\.
• cloneify run compressed (research already done): clone-run\scope.md, research.md, PLAN.md (NOT Codex-grilled), STATE.md. Key architecture decision: outliers self-scored off YouTube Data API ((views / channel avg) * 100), vidIQ optional — app is standalone.
• Two tabs merged into one project (PRD §5.5/§11/§13): her universal-youtube-content-engine skill (250 proven Why Dachshunds packages) is the app's content brain; /roast on a sellable version = RESHAPE -> market version PARKED, personal tool confirmed; old Base44 AmplifAI-Your-Business/ai-youtube-content-engine-system-dashboard marked superseded.
• Roast question answered: its objections (strips human steering, inverted unit economics, commoditized moat) were all cold-market-product objections -> voided by the personal-tool reframe. Survivors: "proven = 250 packages not channel views" caveat (only bites if sold), and "moat = her audience." Offered to re-roast the actual current question (build the app vs just use the skill).

Key files for next session

• c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\YouTube Content Engine\PRD.md — READ FIRST. The unified project; §5.5 (skill as content brain), §8 (phasing), §13 (reconciliation/decision log).
• c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\YouTube Content Engine\FABLE-BUILD-PROMPT.md — paste-ready Fable prompt (Fable 5 / xhigh) + /goal line; build path is a fresh Claude Code session opened in that folder.
• c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\YouTube Content Engine\clone-run\PLAN.md — build-ready spec, 8 milestones (M0–M7), acceptance = "ENGINE V1 PASS". NOT yet Codex-grilled.
• C:\Users\srver\.claude\skills\universal-youtube-content-engine\SKILL.md — the content brain; 7 framework files in its references\ to bundle into the app.
• c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\Sandy Lee AI Content Team\brand-voice.md — becomes the sondraverva profile voice (DRAFT, needs her edit).
• c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\YT - Why Dachshunds Channel\Scripts\Batch 2 - 80 Video_Packages (DONE)\Video_01_Why_Dachshunds_Refuse_To_Walk.md — the 250-package output shape.
• Plan file: none (PRD drove the session, not a plans\ file).
• Memory files touched this session: C:\Users\srver\.claude\projects\c--Users-srver-AIOS-Workspace-aios-starter-kit\memory\project_content_team_engine.md (created, then repointed to the unified project) and ...\memory\MEMORY.md (pointer line updated).

Running state

• Background processes: one stale Bash shell bweifg245 (a find for a screenshot path) — non-load-bearing, ignore/let it exit.
• Dev servers / ports: none (PRD proposes local FastAPI + web UI on localhost:4770 via Start Content Engine.bat, not yet built).
• Open worktrees / branches: none.
• Note: a Playwright MCP browser was opened to https://sleestudio.com/ this session; close it if still open.

Verification — how to confirm things still work

• ls "c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\YouTube Content Engine\" -> PRD.md, FABLE-BUILD-PROMPT.md, clone-run\, handoffs\.
• ls "c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\YouTube Content Engine\clone-run\" -> scope.md, research.md, PLAN.md, STATE.md.
• vidIQ reachable only in Claude via ToolSearch mcp__claude_ai_vidIQ_for_Claude__*; a standalone app cannot use it — YOUTUBE_API_KEY (master .env) is the app's research source.
• Channel identity: python scripts/get_env.py YOUTUBE_CHANNEL_ID -> UCOVr7yl6GJwrtdoWea3lrBw.

Deferred + open questions

• Open (the live decision): how to move on the YouTube Content Engine — (1) build Phase 1 now (Profiles + Trend Scout) [recommended], (2) run grill-me-codex on PLAN.md first, or (3) just review the Fable prompt. Plus: name it (working title "YouTube Content Engine").
• Open: whether to re-run /roast on the actual current question — "build the full multi-channel+faceless app vs. just use the universal-youtube-content-engine skill manually + the Trend Scout." Offered, not yet answered.
• Deferred: sondraverva brand-voice.md still DRAFT — Sondra must correct it before scripts are trustworthy.
• Deferred: standalone sellable/hosted version — PRD keeps v1 private; graduating to a product is out of scope for v1, and would require /app-safety-check + the "channel gets views" proof.

Pick up here

Read c:\Users\srver\AIOS-Workspace\aios-starter-kit\PROJECTS\YouTube Content Engine\PRD.md, then ask Sondra which of the three moves she wants; if she says build, start Phase 1 (Channel Profiles + Trend Scout via YouTube Data API) per PLAN.md M0->M2 as a real usable local tool before adding the rest.