> ## Documentation Index
> Fetch the complete documentation index at: https://docs.synscribe.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Knowledge-gap register

> What Notion sourcing resolved, and what still needs Raymond's call.

The open questions the build surfaced. A Notion-sourcing pass (2026-07-04) closed most of them by
mining pages adjacent to the blank SOPs. What remains below genuinely needs your input or is a
`[DECIDE]` blocked on an in-progress decision (e.g. the pricing rework).

## Resolved this pass (sourced from Notion, now in the pages)

* **1.4 SERP-cliff mechanics** — a cliff is the first of the top 7 results with either no exact
  keyword in the title tag or a low-DA domain; score by depth (1–3 publish now, 4–5 target, 6–7
  defer, 8+ park); only title tags count; aggregators with an exact-match title are locked; cliff
  scores decay, so publish fast. *(SERP Cliff Framework page.)*
* **1.5 2-D keyword framework** — Intent (7 additive labels) by Relevance (single-select, ranked
  `OFFERING > ALTERNATIVE > COMPLEMENTARY > INCUMBENT`); target high-intent by high-relevance; the
  winnability gate is the SERP-cliff read, not a separate gate.
* **3.2 Monthly planning** — runs \~1 week before renewal (review + next-cycle strategy), produces a
  written strategy/keyword doc, front-load generation into week 1, \~10–15% planning buffer.
* **3.3 Weekly planning** — Monday all-account health review against a small standard question set,
  plus a per-client week-by-week deliverables-vs-plan grid.
* **3.5 Content production + DoD** — landing-pages-first, blogs start at 3; Pi generates, AM signs
  off, publishing handoff. Definition of Done is the **15-item must-pass gate** with a quantitative
  table (keyword as first word of H1, meta description 160 chars or fewer, zero broken/competitor
  links, exact-match slug, indexed within \~48h, …).
* **3.6 Quality Gate (ongoing)** — the Month-1 outcome gate stays, plus a per-batch must-pass gate:
  a must-pass fail is not done; same item fails twice, escalate to Raymond (medium), tag
  `quality-gate-debt`.
* **3.10 Press release** — GlobeNewswire (primary) / AccessNewswire (budget) on Notified; Pi or the
  AM drafts, the wire only distributes; classic PR template with keyword-anchored links; honest
  bump caveat (Yahoo/Benzinga placements within \~12h, but click counts are mostly bot/syndication —
  the real signal is backlinks + branded search).
* **3.11 First keyword set** — seed from sales-call transcripts; first wave \~10–20 landing-page
  keywords, first 5 LPs + 3 blogs, landing-page-first.
* **2.2.3 Tag vocabulary** — two systems: freeform pillar/cycle tags (`Core` / `Fringe` /
  `<cycle-month>` / `Alternative` / `from [Client]`) and the structured 2-D labels (7 intent by 4
  relevance). "Alternative" exists on both.
* **2.2.4 Monthly report shape** — a forward-looking before/after positions report, a standard
  section of the Monthly Review.
* **1.1 / 1.7 / 1.12 substance added** — positioning (AI-native, service-first, own-the-outcome vs
  the buyer's three bad options), the real speed-to-rank mechanism + Zero-to-Ranked track record,
  and the four leverage vectors are now on the pages, each with a short confirm/\[DECIDE] note below.

## Still open — needs Raymond

**Part 1 — theory**

* **1.1** — confirm the exact positioning line/claims (substance added; wording is yours to bless).
* **1.2** — the referral-*tracking* rationale (why we track ChatGPT/Perplexity/Claude/Gemini as a
  channel). Related philosophy now in 1.10; confirm.
* **1.7** — confirm the Zero-to-Ranked case-study numbers are OK to publish (top-3 in \~24h; \~67
  clicks to a handful of enterprise leads in two weeks).
* **1.8** — the hub/spoke *equity-flow* model (how much a hub passes, spoke-to-spoke, cross-pillar).
  Not documented anywhere in Notion; the crawler-amplifier framing is in the page.
* **1.9** — the press-release "bump" causal story + magnitude/half-life (3.10 has the honest
  caveat; a clean number would sharpen this).
* **1.10** — confirm the attribution philosophy as canon.
* **1.12** — the concrete "1 AM handles N clients vs a manual operator" ratio (not documented).

**Part 2 — platform specifics** (mostly code-verified; need your convention call)

* **2.0.4** — `/reports/` vs `/memory/`: SOPs say file strategy docs to `/reports/`, code's real
  dir is `/memory/`. Standardize which.
* **2.1.1** bootstrap seeds vs in-chat soul/identity · **2.1.7** the four refiner names (are they
  auto-seeded?) · **2.2.3** operator-facing name of the "keyword content planner" view · **2.3.4**
  code-site export hand-off · **2.3.6** the video-to-LP flow · **2.4.1** "Setting up Blog" / "Refill
  content" Loom scope · **2.5.1** the stale guardrail settings copy · **2.6.2** where the
  user-journey trace / dual contact form are configured.
* **2.8.3** — the DataForSEO BYOK UI surface. The connector mechanism is in code
  (`packages/connectors/src/core-connectors.ts`: defaults to platform creds, flips to a client's own
  `login`/`password`), but there's no DataForSEO field on Settings → Features and no connectors
  settings page in synscribe — so the BYOK entry point may not be shipped in the UI yet. Confirm the
  surface (Settings → Features vs a new connectors page) before we tell operators where to click.

**Part 3 — `[DECIDE]` (blocked on the pricing rework)**

* **Standard-tier output per cycle** (pages/week) — used in 3.5, 3.7, 3.8. Notion shows 1 page per
  credit and a standard tier around 80/month, but a pricing rework is in progress — confirm the
  current number.
* **Whether to name a customer-facing package** (3.7).
* **3.9** the "build strong internal links" SOP + outreach cadence · **3.12** the API/MCP step-by-step
  (its Notion SOP is a client-list stub) · **3.15** any hard agency-partner boundaries · **3.16** the
  capacity numbers (same as the tier `[DECIDE]`).

## How-to library — \[DECIDE] (compliance cluster + GEO)

* **Publish the "evaluator" name?** The new 2.7 pages call the compliance guardrail "the evaluator"
  per your steer; confirm that's the public term (vs "compliance guardrail").
* **A real named regulated example?** 2.7.5 uses a generic fictional fintech. The codebase ships a
  real StraitsX policy fixture — decide whether to publish a named example.
* **2.1.7 refiner names** — resolved by the 2026-07 refiners Loom (`7cf7cd09`): the refiners are **not**
  auto-seeded; Pi creates them on `setup my default refiners`. Canonical UI names are **TLDR Summary**,
  **FAQ Section**, **CTA Banner** (the auto-created set), plus **Product Alignment** added to complete
  the typical four. Confirm this is the standard you want to teach.
* **`/llms.txt` + `.md` mirrors need the hosted deploy** — local `mint dev` 404s them; realizing the
  GEO win (playbook Plays 5/6) requires the Mintlify cloud deploy (ties to the durable-deploy call).

## Cross-cutting

* `references/…` in SOPs is skill-relative, not an operator VFS folder — don't teach it as one.
* `USER.md` referenced in the BOOTSTRAP template but not auto-loaded — status unconfirmed.