One file. Any LLM. Now and future.

Master Website Build Brief

A single, self-contained specification for building any PM Consulting website or page on any large language model. Hand over one file and the model builds to the correct structure and the correct brand.

What is inside

Three parts, in priority order

The brief is built so a model reads it top to bottom and knows exactly what wins when rules meet. House Rules are absolute. The framework governs structure. The supplementary layers only fill gaps.

Part A

House Rules

The non-negotiables that keep output on brand. These override everything below.

  • No em dashes, no emoji (SVG icons only)
  • No FAQ accordions, answers always visible
  • Single client footer line, 50/50 terms
  • Sell the outcome, phases start at 1
Part B

The Framework

The Local SEO Site Generator: the primary authority for how every page is built and scored.

  • 10 capabilities, entity triples, AEO
  • 37 scoring factors, 10 critical gates
  • Entity-first writing, question-led answers
  • Trust-conditional similarity rule
Part C

Supplementary

TJ Robertson and Dennis Yu layers, summarized, with the authority order spelled out.

  • TJ Robertson: GEO, get cited by AI
  • Dennis Yu: content factory, Dollar a Day
  • Never overrides the framework
How to use it

Three steps on any model

1

Open a fresh session on Claude, ChatGPT, Gemini, or whatever you are using.

2

Paste the .md (or upload the .docx) at the start, then say "build to this spec."

3

Give your task. The model now produces the right structure and the right brand, em dashes and emoji excluded, footers and pricing correct.

Behind the brief

Exactly what was done

This brief is the end product of a full pass over the latest GEO research (James Dooley and Sterling Sky, 2026) and a tightening of the whole build system so every surface points to one source of truth.

Framework upgraded

Entity-first writing rule, question-led answers, mandatory pricing block, and a reputation / LLM-seeding layer added.

Similarity gate made smarter

Strict by default on new domains, with a narrow opt-in exception for established brands. Safety kept, finding captured.

Validator now enforces it

Two new scored factors (entity-first opener, pricing signal) with a passing test suite.

One source of truth

A broken config pointer was fixed so terminal, desktop, and Obsidian all route builds to the same framework file.

Portable exports created

The framework and this combined brief in both .md and .docx for use with any model.

One-command rebuild

build-brief.sh reassembles the brief from source any time the rules change. No manual stitching.

This brief is a snapshot assembled from the framework, the house rules, and the two playbooks. If a rule changes, rerun the rebuild script and re-upload the files here to keep this page current.

Control room

Rebuild and source files

Everything that produces this brief, in one place. Edit a source, run the one command, then re-upload the two files to this page to refresh it.

cd ~/Projects/local-seo-toolkit && ./scripts/build-brief.sh
The one command

build-brief.sh

~/Projects/local-seo-toolkit/scripts/build-brief.sh
View
Editable source parts

Part A. House Rules

~/Projects/local-seo-toolkit/docs/brief-parts/part-a-house-rules.md
View

Part B. The Framework (canonical, PRIMARY)

~/Projects/local-seo-toolkit/docs/framework.md
View

Part C. Supplementary layers

~/Projects/local-seo-toolkit/docs/brief-parts/part-c-supplementary.md
View
Enforcement and config

Validator (scoring factors + gates)

~/Projects/local-seo-toolkit/seo_validator/scorer.py
View

Validator tests

~/Projects/local-seo-toolkit/tests/test_entity_pricing_factors.py
View

CLAUDE.md (routes every environment to one file)

~/brain/tools/claude-code/CLAUDE.md
View
Repositories

local-seo-toolkit

framework, validator, brief, build script, exports
Open

brain

CLAUDE.md, memory, cross-Mac config
Open