diff --git a/docs-cms/adr/index.md b/docs-cms/adr/index.md
index 2bf473e8b..3a41f876f 100644
--- a/docs-cms/adr/index.md
+++ b/docs-cms/adr/index.md
@@ -9,6 +9,66 @@ slug: /
Architecture Decision Records document significant architectural choices made in Prism. Each ADR captures the problem context, decision rationale, alternatives considered, and consequencesβcreating a historical record of "why" behind the architecture.
+## Prism Architecture at 10,000 Meters
+
+```mermaid
+flowchart TB
+ subgraph Clients["Client Applications"]
+ App1["Service A"]
+ App2["Service B"]
+ App3["Service C"]
+ end
+
+ subgraph Proxy["Prism Proxy (Rust)"]
+ direction TB
+ API["Unified gRPC/HTTP API"]
+ Router["Request Router"]
+ Auth["Auth & Policy
(mTLS, OAuth2)"]
+ end
+
+ subgraph Patterns["Pattern Layer (Go)"]
+ KV["KeyValue"]
+ PS["PubSub"]
+ Q["Queue"]
+ TS["TimeSeries"]
+ Composite["Composite Patterns
(Outbox, CDC, Claim-Check)"]
+ end
+
+ subgraph Backends["Backend Drivers"]
+ Kafka["Kafka"]
+ Postgres["Postgres"]
+ Redis["Redis"]
+ NATS["NATS"]
+ S3["S3/MinIO"]
+ end
+
+ subgraph Config["Configuration"]
+ NS["Namespace
Requests"]
+ Policy["Platform
Policy"]
+ Registry["Backend
Registry"]
+ end
+
+ Clients --> Proxy
+ API --> Router
+ Router --> Auth
+ Auth --> Patterns
+ Patterns --> Backends
+ Config -.-> Proxy
+ Config -.-> Patterns
+
+ style Clients fill:#e3f2fd,stroke:#1565c0
+ style Proxy fill:#fff8e1,stroke:#f57f17
+ style Patterns fill:#f3e5f5,stroke:#7b1fa2
+ style Backends fill:#e8f5e9,stroke:#2e7d32
+ style Config fill:#fce4ec,stroke:#c2185b
+```
+
+**Key architectural decisions**:
+- **[ADR-001](/adr/adr-001)**: Rust proxy for 10-100x performance over JVM
+- **[ADR-002](/adr/adr-002)**: Client-originated configuration for self-service
+- **[ADR-003](/adr/adr-003)**: Protobuf as single source of truth
+- **[ADR-005](/adr/adr-005)**: Pluggable backend architecture
+
## π― New to Prism? Start Here
If you're exploring Prism's architecture, start with these foundational decisions:
diff --git a/docs/.gitignore b/docs/.gitignore
new file mode 100644
index 000000000..012dd86a2
--- /dev/null
+++ b/docs/.gitignore
@@ -0,0 +1,39 @@
+# docs/.gitignore - GitHub Pages Build Output Directory
+#
+# CRITICAL: This directory contains BUILD OUTPUT ONLY (except img/)
+#
+# - Build output is generated by CI (docusaurus build --out-dir ../docs)
+# - NEVER manually commit HTML/JS/CSS files from this directory
+# - Source documentation is in docusaurus/docs/ and docs-cms/
+#
+# What IS allowed (tracked in git):
+# - .nojekyll (required by GitHub Pages to bypass Jekyll processing)
+# - CNAME (custom domain configuration, if used)
+# - .gitattributes (git attributes for docs/, if needed)
+# - This .gitignore file itself
+# - img/ directory and all images (logos, icons, etc.)
+
+# Ignore HTML pages (all generated)
+*.html
+
+# Ignore generated search index and sitemap
+search-index.json
+sitemap.xml
+
+# Ignore generated directories (docusaurus output)
+/adr/
+/rfc/
+/memos/
+/user-guide/
+/search/
+/netflix/
+/prds/
+/docs/
+
+# Ignore generated JS/CSS in assets/
+/assets/css/
+/assets/js/
+
+# Allow img/ directory and all images
+!/img/
+!/img/**
diff --git a/docs/.nojekyll b/docs/.nojekyll
deleted file mode 100644
index e69de29bb..000000000
diff --git a/docusaurus/docs/intro.md b/docusaurus/docs/intro.md
index 887d293d4..a9812aa88 100644
--- a/docusaurus/docs/intro.md
+++ b/docusaurus/docs/intro.md
@@ -27,21 +27,29 @@ Recent highlights:
**Three layers separate what, how, and where**:
-```
-ββββββββββββββββββββββββββββββββββββββ
-β Client API (What) β Applications use stable APIs
-β KeyValue | PubSub | Queue β
-ββββββββββββββββββββββββββββββββββββββ
- β
-ββββββββββββββββββββββββββββββββββββββ
-β Patterns (How) β Prism applies reliability patterns
-β Outbox | CDC | Claim Check β
-ββββββββββββββββββββββββββββββββββββββ
- β
-ββββββββββββββββββββββββββββββββββββββ
-β Backends (Where) β Data stored in optimal backend
-β Kafka | Postgres | Redis | NATS β
-ββββββββββββββββββββββββββββββββββββββ
+```mermaid
+flowchart TB
+ subgraph L1["Client API (What)"]
+ A1["Applications use stable APIs"]
+ A2["KeyValue | PubSub | Queue"]
+ end
+
+ subgraph L2["Patterns (How)"]
+ B1["Prism applies reliability patterns"]
+ B2["Outbox | CDC | Claim Check"]
+ end
+
+ subgraph L3["Backends (Where)"]
+ C1["Data stored in optimal backend"]
+ C2["Kafka | Postgres | Redis | NATS"]
+ end
+
+ L1 --> L2
+ L2 --> L3
+
+ style L1 fill:#e1f5fe,stroke:#01579b
+ style L2 fill:#fff3e0,stroke:#e65100
+ style L3 fill:#e8f5e9,stroke:#1b5e20
```
**Benefits**:
diff --git a/docusaurus/docs/key.md b/docusaurus/docs/key.md
index ec7d5454d..6cb806cf7 100644
--- a/docusaurus/docs/key.md
+++ b/docusaurus/docs/key.md
@@ -143,6 +143,28 @@ Same pattern works with Redis+NATS+Postgres OR DynamoDB+SNS+SQS by swapping Laye
---
+### Phase 3.5: Configuration Model (10 min)
+
+**Read**: [Configuration Overview](/docs/configuration/overview)
+
+**Why read this**: Understand the 6-layer configuration model that separates user needs from platform implementation.
+
+**The six layers**:
+1. **User Request** (Layer 1) - Applications declare needs via namespace configuration
+2. **Platform Policy** (Layer 2) - Teams have permission levels (Guided, Advanced, Expert)
+3. **Pattern Selection** (Layer 3) - Prism automatically selects patterns based on requirements
+4. **Backend Registry** (Layer 4) - Operators register available backends
+5. **Frontend Registry** (Layer 5) - API protocol bindings (REST, gRPC, GraphQL)
+6. **Runtime Execution** (Layer 6) - Pattern runners and proxy configuration
+
+**Key documents**:
+- [Namespace Configuration](/docs/configuration/namespace-request) - How applications declare needs
+- [Backend Registry](/docs/configuration/backend-registry) - Available backends and capabilities
+- [Slot Matching](/docs/configuration/slot-matching) - How patterns bind to backends
+- [RFC-056: Unified Configuration Model](/rfc/rfc-056) - Complete specification
+
+---
+
### Phase 4: Implementation Roadmap (10 min)
#### [RFC-018: POC Implementation Strategy](/rfc/rfc-018) (7 min)
@@ -196,16 +218,17 @@ Same pattern works with Redis+NATS+Postgres OR DynamoDB+SNS+SQS by swapping Laye
## Your Reading Path
-### New to Prism? (35 min)
+### New to Prism? (45 min)
Follow this sequence to build complete understanding:
1. **Vision** (10 min): [PRD](/prds/prd-001)
2. **Decisions** (15 min): [ADR-001](/adr/adr-001), [ADR-002](/adr/adr-002), [ADR-003](/adr/adr-003), [ADR-004](/adr/adr-004)
3. **Architecture** (10 min): [MEMO-006](/memos/memo-006)
-4. **Development** (5 min): [CLAUDE.md](https://github.com/jrepp/prism-data-layer/blob/main/CLAUDE.md)
+4. **Configuration** (10 min): [Configuration Overview](/docs/configuration/overview)
+5. **Development** (5 min): [CLAUDE.md](https://github.com/jrepp/prism-data-layer/blob/main/CLAUDE.md)
-**After this**: You'll understand Prism's vision, technical foundation, architecture, and development practices.
+**After this**: You'll understand Prism's vision, technical foundation, architecture, configuration model, and development practices.
---
@@ -243,6 +266,7 @@ Understand the decision-making context:
| [PRD](/prds/prd-001) | Product vision | Onboarding, strategic decisions |
| [ADR-001](/adr/adr-001) - [ADR-004](/adr/adr-004) | Core decisions | Understanding technical foundation |
| [MEMO-006](/memos/memo-006) | Three-layer architecture | Designing patterns, understanding backend abstraction |
+| [Configuration Overview](/docs/configuration/overview) | 6-layer config model | Understanding namespace configuration |
| [RFC-018](/rfc/rfc-018) | POC roadmap | Planning implementation work |
| [CLAUDE.md](https://github.com/jrepp/prism-data-layer/blob/main/CLAUDE.md) | Development guide | Daily development, code reviews |
@@ -274,4 +298,4 @@ This guide evolves as the project grows. When adding foundational documents:
*Reading time estimates assume focused reading with note-taking. Skim faster if reviewing familiar concepts.*
-*Last updated: 2025-10-14*
+*Last updated: 2025-11-25*