---
title: Prime Architect
category: product
entity_type: agent
price: $79
canonical: https://forgehouse.ai/agents/prime-architect/
lang: en
hreflang_alt: https://forgehouse.ai/tr/ajanlar/prime-architect/
last_updated: 2026-06-20
---

# Prime Architect

> Read-only 7-dimension system auditor

A read-only chief architect that scans every dimension of a project, surfaces invisible cross-dimension links, and returns a prioritized reality map with the highest-leverage opportunities. It deliberately cannot edit anything: the output is a dispatch plan: findings ranked by impact against effort, each one named to the specialist who should execute it, so the scan ends in decisions, not a report that sits.

## Use cases
- A 360 health scan of a site or system
- Finding what is missing across SEO, speed, data and UX
- Prioritizing opportunities by impact and effort
- Tracing a cross-dimension chain (schema → AEO → traffic → CAC)
- Producing a dispatch blueprint for specialists

## Benefits
- See the whole system, not one symptom at a time
- Spend effort where it actually moves the needle
- Catch hidden links a single-domain review misses
- A prioritized plan instead of a scattered to-do list

## What’s included
- Seven-dimension read-only scans across product, tech, SEO, data, ops, finance and growth
- Cross-dimension link discovery: speed→CWV→quality-score→CPC chains made visible
- Impact×(10−effort) prioritisation with a top-3 opportunity shortlist
- Dispatch plans naming which specialist executes each finding
- Reality maps that separate what exists from what is assumed
- Zero file edits by design: the deliverable is the plan, not the patch

## Who it’s for
Owners who want the full picture before deciding where to invest.

## How it runs
This auditor carries no write tools by design. It sweeps a project across seven dimensions, hunts the cross-dimension chains where one missing schema becomes rising acquisition cost, and returns a dispatch plan instead of touching anything.
1. Reads the system rule index, the rules covering the dimensions under review and recent audit reports first, so it never rediscovers known gaps that are already on the open audit list.
2. Runs a strictly read-only structural scan: project context, package and config files, source tree, file structure and the live site. It deliberately has no write tools, so it cannot mutate anything while auditing.
3. Sweeps the project across seven dimensions (UI/UX, code, SEO, marketing, security, FinOps, psychology) with a key question per dimension, doing the structural pass itself and marking data-heavy dimensions for a specialist to deepen later.
4. Hunts cross-dimension chains, its signature output: a missing schema cascades into AI-search invisibility, organic decline and rising acquisition cost; a slow page hits not just UX but quality score and ad cost. Single-dimension findings are considered incomplete.
5. Prioritizes every opportunity with an impact times ease score and surfaces only the top three, each backed by structural evidence like a grep match, an HTTP status or a file tree count. Anything it could not verify structurally stays labeled as inference.
6. Returns a dispatch plan instead of dispatching: each work item names the right agent or team and carries a complete mission brief (goal, scope, out of scope, done criteria, verification) that the main thread can apply as-is. The agent itself never spawns other agents.

## FAQ
### I suspect my site has problems but cannot name them, is this the right starting point?
That is its exact use case. It runs a read-only scan across seven dimensions: product, tech, SEO, data, ops, finance and growth, and returns a reality map that separates what exists from what is assumed. You come out with named, ranked findings instead of a vague feeling that something is off.

### How does it prioritize findings instead of dumping a 50-item list on me?
Every finding is scored as impact times (10 minus effort), producing a top-3 opportunity shortlist, and the scan looks for cross-dimension chains a single-domain review misses, like speed feeding CWV feeding quality score feeding CPC. The output is a dispatch plan naming which specialist should execute each item.

### Will it fix anything it finds?
No, and that is deliberate. It is read-only by design with zero file edits; the deliverable is the plan, not the patch. Each finding is assigned to the specialist who should execute it, so the audit role and the execution role never blur.

## Price
$79, one-time, no subscription. VAT included.

Related guide: [AI code review and developer workflow](https://forgehouse.ai/guides/ai-code-review/)