diff --git a/CLAUDE.md b/CLAUDE.md
index 8671043..21d412f 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -2,16 +2,26 @@
 
 This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
 
+> **Loading discipline (ADR-0081):** this root file is the always-loaded Layer 0 — repo identity, data-safety/fair-use invariants, and the command quick-start. Deeper, path-conditional material is routed out: forum-scraper internals → `deakins_forums/CLAUDE.md` (loads when editing that subtree); record schemas → `documentation/data-model-reference.md`; structure / podcast-ingest / testing reference → `documentation/repo-reference.md`.
+
 ## Overview
 
 This is a **Team Deakins Podcast and Forum Knowledge Base** project. It contains:
 
 1. **Podcast Episode Downloader & Ingest**: Scripts to download MP3s from RSS feeds and organize them into structured episode directories with metadata and transcripts.
-2. **Forum Scraper (`deakins_forums`)**: A modular, ultra-structured Python package for scraping rogerdeakins.com forums into a searchable JSON-based knowledge base.
-3. **TypeScript UI Layer (`packages/`)**: A Module Federation remote (React/Vite micro-frontend on port 3020 + Express API on port 3021) for exploring podcast and forum data locally. UI components are imported from `@ojfbot/frame-ui-components` (npm `^1.0.1`) (DashboardLayout, ChatShell, ChatMessage, MarkdownMessage, ErrorBoundary, ThreadSidebar, CondensedChat). Exposes a `GET /api/beads` endpoint per the fleet-wide ADR-0016 bead projection contract. Architecture decisions documented in ADR-006 through ADR-009, ADR-0016 (bead projection), and ADR-0030 (shared component architecture).
+2. **Forum Scraper (`deakins_forums`)**: A modular, ultra-structured Python package for scraping rogerdeakins.com forums into a searchable JSON-based knowledge base. (Design + CLI: `deakins_forums/CLAUDE.md`.)
+3. **TypeScript UI Layer (`packages/`)**: A Module Federation remote (React/Vite micro-frontend on port 3020 + Express API on port 3021) for exploring podcast and forum data locally. Exposes `GET /api/beads` per the fleet-wide ADR-0016 bead projection contract. Architecture decisions: ADR-006 through ADR-009, ADR-0016, ADR-0030.
 4. **Standalone Flask UI (`app.py`)**: A single-file dark-theme knowledge browser at `localhost:5050`, reading directly from `downloads/` and `library/forums/`. Zero dependency on the Module Federation stack — used for local debugging against the raw corpus.
+
+## Fair Use & Privacy (must-keep)
+
 All content is for **personal research and educational purposes only** under fair use principles. Commercial use requires explicit permission from copyright holders.
 
+- **Scraped data stays off-git** — the repo is public, but scraped forum content (`library/`), podcast audio (`downloads/`), and exports (`analysis/`) are gitignored and never committed. Only source code and illustrative schema examples are public. See Git Data Policy below.
+- **Personal/educational use only** — no commercial redistribution. Contact rogerdeakins.com / Team Deakins for commercial licensing.
+- Scripts identify research purpose via User-Agent headers; rate limiting is enforced (3s default); respect `robots.txt` and server resources.
+- All content remains property of original copyright holders.
+
 ## Git Data Policy
 
 **What is committed to git:**
@@ -28,50 +38,6 @@ All content is for **personal research and educational purposes only** under fai
 
 **Schema examples vs. actual data:** The `documentation/examples-schemas/` directory is the git-committed representation of the data layer. Each file shows the shape of a record type with fictional illustrative content — not real scraped data. See `documentation/examples-schemas/README.md` for field-level documentation.
 
-## Project Structure
-
-```
-purefoy/
-├── deakins_forums/          # Forum scraper package (structured, MCP-ready)
-│   ├── cli.py              # Command-line interface
-│   ├── pipeline.py         # High-level scraping orchestration
-│   ├── models.py           # Pydantic data models (PostLeaf, TopicLeaf, etc.)
-│   ├── parser_bbpress.py   # HTML parsing logic for bbPress forums
-│   ├── http_client.py      # Rate-limited HTTP client with ETag support
-│   ├── store_json.py       # JSON "leaf" persistence layer
-│   ├── index_sqlite.py     # SQLite FTS5 search index
-│   ├── normalize.py        # Content extraction (links, quotes, blocks)
-│   ├── export.py           # Export utilities (CSV, text, by-author, etc.)
-│   ├── coverage.py         # Coverage tracking for incremental scrapes
-│   ├── query_tracker.py    # Query provenance/lineage tracking
-│   ├── report.py           # Reporting utilities
-│   └── config.py           # Settings with env var overrides
-├── packages/                # TypeScript UI layer (pnpm workspace)
-│   ├── browser-app/        # React/Vite micro-frontend remote (port 3020) — UI via @ojfbot/frame-ui-components (npm ^1.0.1)
-│   ├── browser-automation/ # Visual regression testing via browser automation (screenshot capture, CI integration)
-│   ├── api/                # Express API over flat JSON + SQLite (port 3021) — includes GET /api/beads (ADR-0016)
-│   └── shared/             # @purefoy/shared — generated OpenAPI schema + API type contracts (incl. GoalManifest, ReviewProgress)
-├── app.py                   # Standalone Flask UI — dark-theme knowledge browser (port 5050)
-├── download_episodes.py     # Podcast RSS feed downloader (standalone)
-├── ingest_teamdeakins_downloads.py  # Episode ingest with metadata & transcripts
-├── scrape_forum.py          # Legacy forum scraper (standalone)
-├── library/                 # Output directory for scraped data
-│   └── forums/             # Forum JSON leafs organized by type
-│       ├── posts/          # Individual post JSON files
-│       ├── topics/         # Topic index files
-│       ├── forums/         # Forum metadata files
-│       └── _site/          # Sqlite index, HTTP state, coverage
-├── downloads/               # Podcast episode directories (348/348 episodes transcribed)
-│   └── S02E176__2026-01-14__guest-name__libsyn_abc123/
-│       ├── audio.mp3
-│       ├── metadata.json
-│       └── transcript/
-├── analysis/                # Export output (CSV, text dumps, stats)
-├── package.json             # pnpm workspace root
-└── pyproject.toml           # Python package metadata
-```
-```
-
 ## Deployment
 
 **NEVER deploy directly to production** via CLI (`vercel deploy --prod`, `aws s3 sync` to prod paths, etc.).
@@ -96,57 +62,6 @@ pip install -r requirements.txt
 pip install -e .
 ```
 
-### Podcast Episode Management
-
-```bash
-# Download MP3s from RSS feed
-.venv/bin/python3 download_episodes.py
-
-# Organize downloads into episode directories (dry-run)
-.venv/bin/python3 ingest_teamdeakins_downloads.py \
-  --rss "https://rss.libsyn.com/shows/265448/destinations/2018942.xml" \
-  --downloads "./downloads" \
-  --dry-run
-
-# Actually ingest with transcript fetching
-.venv/bin/python3 ingest_teamdeakins_downloads.py \
-  --rss "https://rss.libsyn.com/shows/265448/destinations/2018942.xml" \
-  --downloads "./downloads" \
-  --fill-transcripts \
-  --tapesearch
-```
-
-### Forum Scraping (deakins_forums package)
-
-```bash
-# Using the CLI entry point (if installed)
-deakins-forums scrape-all --max-topics 1000 --build-index
-deakins-forums scrape-forum team-deakins --max-pages 20 --build-index
-deakins-forums scrape-topic https://rogerdeakins.com/forums/topic/...
-
-# Or using python -m
-python -m deakins_forums.cli scrape-forum team-deakins --max-pages 20
-python -m deakins_forums.cli build-index
-python -m deakins_forums.cli search "lighting techniques" --limit 20
-python -m deakins_forums.cli stats
-
-# Export data
-python -m deakins_forums.cli export all-text -o analysis/all_posts.txt
-python -m deakins_forums.cli export csv -o analysis/posts.csv
-python -m deakins_forums.cli export roger-only -o analysis/roger_posts.txt
-
-# Query tracking (provenance)
-python -m deakins_forums.cli scrape-forum team-deakins \
-  --query-name "Research natural lighting techniques" \
-  --querier-role "Cinematographer" \
-  --querier-department "Camera" \
-  --query-context "TV miniseries prep" \
-  --query-intent "Find Team Deakins insights on natural light"
-
-python -m deakins_forums.cli provenance --topic "team-deakins__topic-slug"
-python -m deakins_forums.cli coverage --format full
-```
-
 ### Code Quality (optional-dependencies)
 
 ```bash
@@ -160,218 +75,11 @@ ruff check .
 pytest
 ```
 
-## Architecture
-
-### Forum Scraper Design
-
-The `deakins_forums` package follows a **composable pipeline architecture**:
-
-1. **HttpClient** (`http_client.py`): Rate-limited HTTP layer with conditional GET (ETag/Last-Modified) support for efficient incremental updates.
-
-2. **Parser** (`parser_bbpress.py`): BeautifulSoup-based HTML extraction for bbPress forum structure. Extracts forums index, forum pages, topic pages, and posts.
-
-3. **Models** (`models.py`): Pydantic models defining the "leaf" JSON structure (v2 schema):
-   - `PostLeaf`: Individual post with content, author, timestamps, structured blocks, quotes, links, media, **plus threading fields** (post_type, parent_post_id, parent_type, position)
-   - `TopicLeaf`: Topic metadata + post ID index (posts are separate files) **plus reply tree** (recursive nested structure showing conversation hierarchy with thread statistics)
-   - `ForumLeaf`: Forum metadata + topic reference list
-   - `PostType` enum: Distinguishes "topic" (starter) from "reply" posts
-   - All include `Provenance` (source URL, scrape timestamp, HTTP headers) and `Integrity` (content hash for change detection, parser_version tracks schema)
-
-4. **Pipeline** (`pipeline.py`): High-level orchestration gluing fetch → parse → normalize → store. Handles pagination, deduplication, incremental refresh.
-
-5. **Storage** (`store_json.py`): Persists structured JSON "leafs" with deterministic file paths (e.g., `posts/{post_id}.json`). Enables version control and incremental updates.
-
-6. **Index** (`index_sqlite.py`): SQLite FTS5 full-text search index for fast queries across posts. Rebuilt from JSON leafs on demand.
-
-7. **Normalize** (`normalize.py`): Extracts structured content elements (links, media, quotes, content blocks) from raw HTML/text for agent-friendly consumption.
-
-8. **Query Tracking** (`query_tracker.py`): Records provenance of each scrape query (who, why, when, what was accessed) for research auditability.
-
-### Key Design Principles
-
-- **Incremental & Resumable**: HTTP ETag caching, content hashing, and visited URL tracking allow efficient re-scraping
-- **Deterministic Paths**: Slug-based file naming enables easy navigation and version control
-- **Agent-Friendly**: Ultra-structured JSON with normalized content blocks, quotes, links for downstream AI/MCP tools
-- **Provenance-First**: Every leaf includes source URL, scrape timestamp, HTTP headers, content hash
-- **Rate-Limited & Respectful**: Configurable delays (default 3s), User-Agent identification, robots.txt compliance
-
-### Episode Ingest Design
-
-The `ingest_teamdeakins_downloads.py` script:
-- Parses RSS XML to extract episode metadata (title, date, duration, iTunes tags, description)
-- Matches downloaded MP3s to RSS entries via URL
-- Creates structured episode directories with naming: `S{season}E{episode}__{date}__{guest}__{source_id}/`
-- Generates `metadata.json` with comprehensive provenance
-- Creates `transcript/` scaffold with `sources.json` and `transcript.txt` (placeholder or fetched)
-- Supports Podcasting 2.0 `<podcast:transcript>` URLs and optional Tapesearch fallback
-
-## Environment Variables
-
-Configure via environment variables (all optional):
-
-```bash
-export DEAKINS_BASE_URL="https://rogerdeakins.com"
-export DEAKINS_USER_AGENT="Your Custom User Agent"
-export DEAKINS_DELAY_S="3.0"          # Seconds between requests
-export DEAKINS_TIMEOUT_S="30"         # Request timeout
-export DEAKINS_MAX_RETRIES="3"
-export DEAKINS_OUT_DIR="./library/forums"
-```
-
-## Data Model Reference
-
-### PostLeaf Structure (v2 Schema)
-
-**NEW in v2:** Threading support with parent-child relationships and post types.
-
-```json
-{
-  "ids": {
-    "post_id": "175763",
-    "forum_slug": "team-deakins",
-    "topic_slug": "lighting-discussion",
-    "reply_permalink": "https://...",
-    "parent_post_id": "175760",
-    "parent_type": "topic",
-    "position": 1
-  },
-  "post_type": "reply",
-  "author": {
-    "display_name": "Roger A. Deakins",
-    "role": "Keymaster"
-  },
-  "timestamps": {
-    "raw": "January 18, 2023 at 3:32 pm #175763",
-    "parsed_iso": "2023-01-18T15:32:00",
-    "parse_confidence": "medium"
-  },
-  "content_text": "...",
-  "content_html": "...",
-  "blocks": [...],
-  "quotes": [...],
-  "links": [...],
-  "media": [...],
-  "provenance": {
-    "source_url": "...",
-    "scraped_at": "2026-01-15T...",
-    "http": {"etag": "...", "last_modified": "..."}
-  },
-  "integrity": {
-    "content_hash": "sha256:...",
-    "parser_version": "bbpress-v2"
-  }
-}
-```
-
-**Threading Fields:**
-- `post_type`: "topic" (starter) or "reply"
-- `parent_post_id`: ID of parent post (null for topics)
-- `parent_type`: "topic" or "reply" (null for topics)
-- `position`: Sequential position in reply list (1-indexed)
-
-### TopicLeaf Structure (v2 Schema)
-
-**NEW in v2:** Hierarchical reply tree with thread statistics.
-
-```json
-{
-  "topic_url": "https://...",
-  "topic_slug": "lighting-discussion",
-  "title": "Lighting Discussion",
-  "post_ids": ["175760", "175763", "175764"],
-  "reply_tree": {
-    "post_id": "175760",
-    "post_type": "topic",
-    "author": "John Doe",
-    "timestamp": "2023-01-18T15:00:00",
-    "children": [
-      {
-        "post_id": "175763",
-        "post_type": "reply",
-        "author": "Roger A. Deakins",
-        "timestamp": "2023-01-18T15:32:00",
-        "children": []
-      },
-      {
-        "post_id": "175764",
-        "post_type": "reply",
-        "author": "Jane Smith",
-        "timestamp": "2023-01-18T16:00:00",
-        "children": []
-      }
-    ]
-  },
-  "reply_count": 2,
-  "max_depth": 1,
-  "provenance": {...},
-  "integrity": {
-    "content_hash": "sha256:...",
-    "parser_version": "bbpress-v2"
-  }
-}
-```
-
-**Reply Tree Fields:**
-- `reply_tree`: Recursive nested structure showing conversation flow
-- `reply_count`: Total number of replies (excluding topic starter)
-- `max_depth`: Maximum nesting depth (0 = no replies, 1 = direct replies only)
-- `post_ids`: Flat list for backward compatibility
-
-**Migration Note:** Use `python -m deakins_forums.migrate --rebuild-trees` to rebuild reply trees for existing data.
-
-### Episode metadata.json Structure
-
-```json
-{
-  "title": "SEASON 2 - EPISODE 176 - Chris Lowe...",
-  "guid": "...",
-  "pubDate_iso": "2026-01-14",
-  "itunes": {
-    "season": 2,
-    "episode": 176,
-    "duration_s": 5400,
-    "summary": "..."
-  },
-  "rss_description": "...",
-  "rss_content_encoded": "...",
-  "audio": {
-    "url": "https://...",
-    "mime": "audio/mpeg",
-    "size_bytes": 123456789,
-    "bitrate_kbps": 128,
-    "duration_s": 5400,
-    "format": "MP3"
-  },
-  "provenance": {
-    "rss_url": "...",
-    "scraped_at": "2026-01-15T...",
-    "script_version": "ingest-v2"
-  }
-}
-```
-
-## Testing Notes
-
-- **TypeScript layer**: CI runs TypeScript type-check, test, and codegen drift guard jobs (see `packages/`). Fleet-wide visual regression tests run via the `browser-automation` package (screenshots capture the composed shell, not standalone apps; waits for a Frame sentinel rather than `networkidle`).
-- **Python layer**: Manual testing via CLI commands and dry-run flags; no formal Python test suite yet
-- Incremental updates can be verified by running scrapes twice and checking for "Not modified" messages
-
-## Important Fair Use Reminders
-
-- **Personal/educational use only** - no commercial redistribution
-- **Keep repository PRIVATE** - scraped forum data is included for research
-- Scripts include User-Agent headers identifying research purpose
-- Rate limiting is enforced (3s default delay between requests)
-- Respect robots.txt and server resources
-- All content remains property of original copyright holders
-- Contact rogerdeakins.com / Team Deakins for commercial licensing
-
-## Git Repository Notes
+**Forum scraping commands** (scrape, index, search, export, provenance) → `deakins_forums/CLAUDE.md`.
+**Podcast download & ingest commands** → `documentation/repo-reference.md`.
 
-This repository previously included scraped forum data (`library/forums/`) but that directory is now gitignored (see Git Data Policy above). On-disk data is for private research use:
-- Forum content is publicly accessible (not behind authentication)
-- Structured JSON format is git-friendly
-- Enables incremental updates and change tracking
-- **Keep this repository PRIVATE** - do not make public
+## Architecture & data model
 
-Audio files (`.mp3`) and episode metadata are excluded via `.gitignore`.
+- **Forum scraper pipeline + design principles** → `deakins_forums/CLAUDE.md`
+- **Record schemas** (PostLeaf, TopicLeaf, Episode metadata; authoritative source `deakins_forums/models.py`) → `documentation/data-model-reference.md`
+- **Project structure, episode-ingest design, testing notes** → `documentation/repo-reference.md`
diff --git a/GIT_READY.md b/GIT_READY.md
index ab9248a..793ccf8 100644
--- a/GIT_READY.md
+++ b/GIT_READY.md
@@ -104,7 +104,7 @@ purefoy/
 - `PROJECT_STRUCTURE.md` - Structure guide
 - `GIT_READY.md` - This file
 
-**Data (Private Repo):**
+**Data (on-disk only — gitignored, not committed):**
 - `library/forums/` - 3,075 post JSONs, 691 topic JSONs (24MB)
 - `library/articles/` - Sample article JSONs
 
@@ -181,20 +181,18 @@ purefoy/
 ## Git Repository Configuration
 
 ### Repository Type
-**Recommended:** Private repository
+**Adopted:** Public repository, code only — scraped data kept off-git.
+
+**Configuration:**
+1. `library/`, `downloads/`, and `analysis/` are in `.gitignore`
+2. Only source code, documentation, and illustrative schema examples are committed
+3. All ~24MB of scraped data stays on-disk (never committed)
+4. Content copyright disclaimer in README / CLAUDE.md
 
 **Reasoning:**
 - Forum content is publicly accessible but should not be redistributed
 - Personal research and educational use only
-- Includes 24MB of scraped JSON data (library/)
-- Respects copyright while enabling version control
-
-### Alternative: Public Code Only
-If making code public:
-1. Add `library/` to `.gitignore`
-2. Only commit source code and documentation
-3. Exclude all scraped data
-4. Add prominent disclaimer about content copyright
+- Respects copyright while keeping the codebase open
 
 ---
 
@@ -282,7 +280,7 @@ git push -u origin main
 ## Next Steps After Git Init
 
 1. **Create Remote Repository**
-   - GitHub, GitLab, or Gitea (private)
+   - GitHub, GitLab, or Gitea
    - Add remote: `git remote add origin <url>`
 
 2. **Push Initial Commit**
diff --git a/PROJECT_STRUCTURE.md b/PROJECT_STRUCTURE.md
index 2ba0148..a03f3a2 100644
--- a/PROJECT_STRUCTURE.md
+++ b/PROJECT_STRUCTURE.md
@@ -78,7 +78,7 @@ __pycache__/               # Python bytecode
 - **`deakins_articles/`**: Articles scraper for rogerdeakins.com/articles
 
 ### Data Storage
-- **`library/`**: Primary data storage with JSON "leafs" (TRACKED in git for private repo)
+- **`library/`**: Primary data storage with JSON "leafs" (gitignored — on-disk only; public repo keeps scraped data off-git)
   - Structured, version-controllable JSON files
   - Enables incremental updates via content hashing
   - Ready for AI/MCP consumption
diff --git a/README.md b/README.md
index 79640ca..cd68e1d 100644
--- a/README.md
+++ b/README.md
@@ -126,7 +126,7 @@ Scrapes technical articles from rogerdeakins.com/articles:
 ### Installation
 
 ```bash
-# Clone repository (private repo assumed)
+# Clone repository
 git clone <repo-url>
 cd purefoy
 
@@ -312,9 +312,13 @@ http_state.json            # HTTP cache (ephemeral)
 
 **What IS tracked:**
 - ✅ All source code (`deakins_forums/`, `deakins_articles/`)
-- ✅ Library data (`library/`) - Structured JSON for private repo
 - ✅ Documentation and guides
 - ✅ Configuration files
+- ✅ Illustrative schema examples (`documentation/examples-schemas/`)
+
+**What is NOT tracked (gitignored, on-disk only):**
+- ❌ Scraped forum content (`library/`) — public repo, so data stays off-git
+- ❌ Podcast audio + episode dirs (`downloads/`), exports (`analysis/`)
 
 ---
 
@@ -364,7 +368,7 @@ See [CLAUDE.md](CLAUDE.md) for detailed architecture documentation.
 
 **Permitted Use:**
 - ✅ Personal research and education
-- ✅ Private knowledge base for learning
+- ✅ Personal knowledge base for learning
 - ✅ Academic study of cinematography
 
 **Prohibited Use:**
@@ -377,21 +381,14 @@ See [CLAUDE.md](CLAUDE.md) for detailed architecture documentation.
 
 ---
 
-### Git Repository Privacy
-
-**⚠️ This MUST be a PRIVATE repository**
-
-The project is configured for a private git repository:
+### Git Data Policy
 
-- `library/` directory contains JSON data (forum content) tracked in the private repo
-- Forum content is publicly accessible but should not be publicly redistributed
-- Use for personal/educational research only
-- Do not push to public GitHub/GitLab
+This is a **public repository** — scraped data is kept off-git, not behind repo visibility.
 
-**If making code public:**
-- Exclude `library/` directory entirely
-- Include only source code and documentation
-- Add clear disclaimers about content copyright
+- `library/` (forum content), `downloads/` (podcast audio), and `analysis/` (exports) are **gitignored** and never committed.
+- Only source code, documentation, and illustrative schema examples (`documentation/examples-schemas/`) are public.
+- Forum content is publicly accessible but should not be redistributed; use for personal/educational research only.
+- Content remains copyright of the original holders — see Copyright & Usage above.
 
 ---
 
diff --git a/deakins_forums/CLAUDE.md b/deakins_forums/CLAUDE.md
new file mode 100644
index 0000000..33ce83e
--- /dev/null
+++ b/deakins_forums/CLAUDE.md
@@ -0,0 +1,83 @@
+# deakins_forums — scraper subtree guidance
+
+Path-conditional guidance (ADR-0081 Layer 1): this file loads when you edit the
+`deakins_forums/` forum-scraper package. Repo-wide policy lives in the root `CLAUDE.md`;
+the leaf/record schemas live in `documentation/data-model-reference.md` (authoritative
+source: `deakins_forums/models.py`).
+
+## Forum Scraper Design
+
+The `deakins_forums` package follows a **composable pipeline architecture**:
+
+1. **HttpClient** (`http_client.py`): Rate-limited HTTP layer with conditional GET (ETag/Last-Modified) support for efficient incremental updates.
+
+2. **Parser** (`parser_bbpress.py`): BeautifulSoup-based HTML extraction for bbPress forum structure. Extracts forums index, forum pages, topic pages, and posts.
+
+3. **Models** (`models.py`): Pydantic models defining the "leaf" JSON structure (v2 schema):
+   - `PostLeaf`: Individual post with content, author, timestamps, structured blocks, quotes, links, media, **plus threading fields** (post_type, parent_post_id, parent_type, position)
+   - `TopicLeaf`: Topic metadata + post ID index (posts are separate files) **plus reply tree** (recursive nested structure showing conversation hierarchy with thread statistics)
+   - `ForumLeaf`: Forum metadata + topic reference list
+   - `PostType` enum: Distinguishes "topic" (starter) from "reply" posts
+   - All include `Provenance` (source URL, scrape timestamp, HTTP headers) and `Integrity` (content hash for change detection, parser_version tracks schema)
+
+4. **Pipeline** (`pipeline.py`): High-level orchestration gluing fetch → parse → normalize → store. Handles pagination, deduplication, incremental refresh.
+
+5. **Storage** (`store_json.py`): Persists structured JSON "leafs" with deterministic file paths (e.g., `posts/{post_id}.json`). Enables version control and incremental updates.
+
+6. **Index** (`index_sqlite.py`): SQLite FTS5 full-text search index for fast queries across posts. Rebuilt from JSON leafs on demand.
+
+7. **Normalize** (`normalize.py`): Extracts structured content elements (links, media, quotes, content blocks) from raw HTML/text for agent-friendly consumption.
+
+8. **Query Tracking** (`query_tracker.py`): Records provenance of each scrape query (who, why, when, what was accessed) for research auditability.
+
+### Key Design Principles
+
+- **Incremental & Resumable**: HTTP ETag caching, content hashing, and visited URL tracking allow efficient re-scraping
+- **Deterministic Paths**: Slug-based file naming enables easy navigation and version control
+- **Agent-Friendly**: Ultra-structured JSON with normalized content blocks, quotes, links for downstream AI/MCP tools
+- **Provenance-First**: Every leaf includes source URL, scrape timestamp, HTTP headers, content hash
+- **Rate-Limited & Respectful**: Configurable delays (default 3s), User-Agent identification, robots.txt compliance
+
+## Environment Variables
+
+Configure via environment variables (all optional):
+
+```bash
+export DEAKINS_BASE_URL="https://rogerdeakins.com"
+export DEAKINS_USER_AGENT="Your Custom User Agent"
+export DEAKINS_DELAY_S="3.0"          # Seconds between requests
+export DEAKINS_TIMEOUT_S="30"         # Request timeout
+export DEAKINS_MAX_RETRIES="3"
+export DEAKINS_OUT_DIR="./library/forums"
+```
+
+## Forum Scraping commands
+
+```bash
+# Using the CLI entry point (if installed)
+deakins-forums scrape-all --max-topics 1000 --build-index
+deakins-forums scrape-forum team-deakins --max-pages 20 --build-index
+deakins-forums scrape-topic https://rogerdeakins.com/forums/topic/...
+
+# Or using python -m
+python -m deakins_forums.cli scrape-forum team-deakins --max-pages 20
+python -m deakins_forums.cli build-index
+python -m deakins_forums.cli search "lighting techniques" --limit 20
+python -m deakins_forums.cli stats
+
+# Export data
+python -m deakins_forums.cli export all-text -o analysis/all_posts.txt
+python -m deakins_forums.cli export csv -o analysis/posts.csv
+python -m deakins_forums.cli export roger-only -o analysis/roger_posts.txt
+
+# Query tracking (provenance)
+python -m deakins_forums.cli scrape-forum team-deakins \
+  --query-name "Research natural lighting techniques" \
+  --querier-role "Cinematographer" \
+  --querier-department "Camera" \
+  --query-context "TV miniseries prep" \
+  --query-intent "Find Team Deakins insights on natural light"
+
+python -m deakins_forums.cli provenance --topic "team-deakins__topic-slug"
+python -m deakins_forums.cli coverage --format full
+```
diff --git a/documentation/GIT_INIT_CHECKLIST.md b/documentation/GIT_INIT_CHECKLIST.md
index 9397326..c2b02c9 100644
--- a/documentation/GIT_INIT_CHECKLIST.md
+++ b/documentation/GIT_INIT_CHECKLIST.md
@@ -94,18 +94,13 @@ git ls-files --others --exclude-standard
 **Schema Examples:**
 - `schema_examples/**/*`
 
-**Forum Scraped Data (for private repo):**
-- `library/forums/posts/*.json`
-- `library/forums/topics/*.json`
-- `library/forums/forums/*.json`
-- `library/forums/_site/forums_index.json`
-
 **Placeholder files:**
 - `downloads/.gitkeep`
 - `analysis/.gitkeep`
 
 ### Files that should NOT be tracked:
 
+❌ `library/**` (scraped forum/article data — public repo keeps it off-git)
 ❌ `downloads/S*/**` (episode directories with audio)
 ❌ `*.mp3`, `*.m4a`, `*.wav` (audio files)
 ❌ `*.sqlite`, `*.db` (search index - can be rebuilt)
@@ -124,7 +119,7 @@ git add .
 git diff --cached --name-only | grep -E "(\.mp3|\.sqlite|\.env|downloads/S)"
 
 # If the above command returns nothing, proceed
-# (Note: library/forums IS tracked - that's expected)
+# (Note: library/ scraped data is gitignored — NOT committed)
 git commit -m "Initial commit: Team Deakins knowledge base toolkit
 
 - Forum scraper with structured JSON storage
@@ -132,7 +127,7 @@ git commit -m "Initial commit: Team Deakins knowledge base toolkit
 - Schema examples and documentation
 - Comprehensive .gitignore for copyright/privacy
 
-Forum data included for private research use.
+Public repo, code only: scraped forum data is gitignored (on-disk).
 Audio files and SQLite indices excluded."
 ```
 
@@ -148,8 +143,8 @@ du -sh .git
 git ls-files | wc -l
 git ls-files | head -20
 
-# Verify forum data IS tracked (should find files)
-git ls-files | grep "library/forums" && echo "✓ Forum data is tracked" || echo "Warning: No forum data found"
+# Verify scraped forum data is NOT tracked (gitignored — should find nothing)
+git ls-files | grep "library/" && echo "ERROR: scraped data tracked!" || echo "✓ Forum data correctly gitignored"
 
 # Verify excluded files are NOT tracked
 git ls-files | grep -E "(downloads/S|\.mp3|\.sqlite|analysis/.*\.csv)" && echo "ERROR: Excluded files tracked!" || echo "✓ Excluded files not tracked"
@@ -235,7 +230,7 @@ git push origin --force --all
 ✅ Documentation tracked
 ✅ Schema examples tracked
 ✅ Configuration tracked
-✅ **Forum scraped data tracked** (for private repo)
+❌ **Forum scraped data excluded** (`library/` gitignored — public repo, off-git)
 ❌ Audio files excluded (.mp3, .m4a, etc.)
 ❌ Episode metadata excluded (downloads/)
 ❌ SQLite indices excluded (can rebuild)
@@ -243,7 +238,7 @@ git push origin --force --all
 ❌ Sensitive data excluded
 
 This ensures:
-- **Privacy**: Keep repository **PRIVATE** (contains scraped forum data)
+- **Privacy**: scraped forum data stays off-git via `.gitignore` (repo is public, code only)
 - **Security**: No credentials or API keys
 - **Copyright**: For personal/educational use only
 - **Size**: Audio files excluded to manage repository size
diff --git a/documentation/data-model-reference.md b/documentation/data-model-reference.md
new file mode 100644
index 0000000..00364cd
--- /dev/null
+++ b/documentation/data-model-reference.md
@@ -0,0 +1,137 @@
+# Data Model Reference
+
+Deep schema reference for the forum/podcast knowledge base (ADR-0081 Layer 2 — task
+reference, not always-loaded). **Authoritative source:** `deakins_forums/models.py`
+(Pydantic). Git-committed illustrative examples: `documentation/examples-schemas/`
+(see its `README.md` for field-level docs). This file mirrors those shapes for quick
+reference when working on the data layer.
+
+## PostLeaf Structure (v2 Schema)
+
+**NEW in v2:** Threading support with parent-child relationships and post types.
+
+```json
+{
+  "ids": {
+    "post_id": "175763",
+    "forum_slug": "team-deakins",
+    "topic_slug": "lighting-discussion",
+    "reply_permalink": "https://...",
+    "parent_post_id": "175760",
+    "parent_type": "topic",
+    "position": 1
+  },
+  "post_type": "reply",
+  "author": {
+    "display_name": "Roger A. Deakins",
+    "role": "Keymaster"
+  },
+  "timestamps": {
+    "raw": "January 18, 2023 at 3:32 pm #175763",
+    "parsed_iso": "2023-01-18T15:32:00",
+    "parse_confidence": "medium"
+  },
+  "content_text": "...",
+  "content_html": "...",
+  "blocks": [...],
+  "quotes": [...],
+  "links": [...],
+  "media": [...],
+  "provenance": {
+    "source_url": "...",
+    "scraped_at": "2026-01-15T...",
+    "http": {"etag": "...", "last_modified": "..."}
+  },
+  "integrity": {
+    "content_hash": "sha256:...",
+    "parser_version": "bbpress-v2"
+  }
+}
+```
+
+**Threading Fields:**
+- `post_type`: "topic" (starter) or "reply"
+- `parent_post_id`: ID of parent post (null for topics)
+- `parent_type`: "topic" or "reply" (null for topics)
+- `position`: Sequential position in reply list (1-indexed)
+
+## TopicLeaf Structure (v2 Schema)
+
+**NEW in v2:** Hierarchical reply tree with thread statistics.
+
+```json
+{
+  "topic_url": "https://...",
+  "topic_slug": "lighting-discussion",
+  "title": "Lighting Discussion",
+  "post_ids": ["175760", "175763", "175764"],
+  "reply_tree": {
+    "post_id": "175760",
+    "post_type": "topic",
+    "author": "John Doe",
+    "timestamp": "2023-01-18T15:00:00",
+    "children": [
+      {
+        "post_id": "175763",
+        "post_type": "reply",
+        "author": "Roger A. Deakins",
+        "timestamp": "2023-01-18T15:32:00",
+        "children": []
+      },
+      {
+        "post_id": "175764",
+        "post_type": "reply",
+        "author": "Jane Smith",
+        "timestamp": "2023-01-18T16:00:00",
+        "children": []
+      }
+    ]
+  },
+  "reply_count": 2,
+  "max_depth": 1,
+  "provenance": {...},
+  "integrity": {
+    "content_hash": "sha256:...",
+    "parser_version": "bbpress-v2"
+  }
+}
+```
+
+**Reply Tree Fields:**
+- `reply_tree`: Recursive nested structure showing conversation flow
+- `reply_count`: Total number of replies (excluding topic starter)
+- `max_depth`: Maximum nesting depth (0 = no replies, 1 = direct replies only)
+- `post_ids`: Flat list for backward compatibility
+
+**Migration Note:** Use `python -m deakins_forums.migrate --rebuild-trees` to rebuild reply trees for existing data.
+
+## Episode metadata.json Structure
+
+```json
+{
+  "title": "SEASON 2 - EPISODE 176 - Chris Lowe...",
+  "guid": "...",
+  "pubDate_iso": "2026-01-14",
+  "itunes": {
+    "season": 2,
+    "episode": 176,
+    "duration_s": 5400,
+    "summary": "..."
+  },
+  "rss_description": "...",
+  "rss_content_encoded": "...",
+  "audio": {
+    "url": "https://...",
+    "mime": "audio/mpeg",
+    "size_bytes": 123456789,
+    "bitrate_kbps": 128,
+    "duration_s": 5400,
+    "format": "MP3"
+  },
+  "provenance": {
+    "rss_url": "...",
+    "scraped_at": "2026-01-15T...",
+    "script_version": "ingest-v2"
+  }
+}
+```
diff --git a/documentation/development/PROJECT_STATUS.md b/documentation/development/PROJECT_STATUS.md
index 8702979..9b02574 100644
--- a/documentation/development/PROJECT_STATUS.md
+++ b/documentation/development/PROJECT_STATUS.md
@@ -63,12 +63,12 @@ purefoy/
     └── README.md
 ```
 
-### Will be included in git ✅ (Forum Data)
+### Excluded from git ❌ (Forum Data — gitignored, on-disk only)
 
-**Note: This is for a PRIVATE repository**
+The repo is public, so scraped forum data is never committed — it lives on-disk only:
 
 ```
-├── library/forums/                     # Scraped forum data
+├── library/forums/                     # Scraped forum data (gitignored)
 │   ├── posts/*.json                   # Individual post files
 │   ├── topics/*.json                  # Topic metadata files
 │   ├── forums/*.json                  # Forum metadata files
@@ -165,20 +165,19 @@ git ls-files | head -20
 
 ## Privacy & Copyright Protection
 
-**⚠️ IMPORTANT: Keep this repository PRIVATE**
+**This is a public repository — scraped data is kept off-git via `.gitignore`.**
 
+✅ **Forum data excluded** - `library/` gitignored; scraped JSON stays on-disk only
 ✅ **Audio files excluded** - No .mp3, .m4a, etc. in git
-✅ **Forum data INCLUDED** - JSON forum data tracked for private research
 ✅ **Episode metadata excluded** - No episode directories tracked
 ✅ **SQLite indices excluded** - Can be rebuilt from JSON
 ✅ **Analysis exports excluded** - Can be regenerated
 ✅ **Sensitive data excluded** - No .env, credentials, or API keys in git
 
-**Why include forum data?**
-- Private repository for personal/educational research
-- Forum content is publicly accessible (not behind authentication)
-- Structured JSON is git-friendly
-- Enables incremental updates and version control
+**Why exclude forum data?**
+- Public repository — scraped content should not be redistributed
+- Keeps the codebase open while respecting copyright
+- Data is regenerable on-disk via the scraper; no need to commit it
 
 ## If You Need Help
 
@@ -192,15 +191,14 @@ git ls-files | head -20
 🎉 **Project is ready for git initialization!**
 
 Your .gitignore is properly configured to:
-- **Include forum data** for private research use
+- **Exclude forum data** (`library/`) — kept off-git on a public repo
 - **Exclude audio files** to manage repository size
 - **Exclude SQLite indices** (can be rebuilt)
 - **Exclude episode metadata** (contains audio files)
 - **Secure credentials** (no sensitive data)
 
-**CRITICAL: Keep repository PRIVATE**
-- Forum data is included for research purposes
-- Personal/educational use only
-- Do not make this repository public
+**Data handling (public repo):**
+- Scraped forum data is gitignored — never committed
+- Personal/educational use only; do not redistribute scraped content
 
 Proceed with `git init` when ready.
diff --git a/documentation/repo-reference.md b/documentation/repo-reference.md
new file mode 100644
index 0000000..50095c8
--- /dev/null
+++ b/documentation/repo-reference.md
@@ -0,0 +1,85 @@
+# Repo Reference
+
+Deep reference for the purefoy knowledge base (ADR-0081 Layer 2 — task reference, not
+always-loaded). Repo-wide policy and the command quick-start live in the root `CLAUDE.md`;
+the forum-scraper internals live in `deakins_forums/CLAUDE.md`; record schemas in
+`documentation/data-model-reference.md`.
+
+## Project Structure
+
+```
+purefoy/
+├── deakins_forums/          # Forum scraper package (structured, MCP-ready)
+│   ├── cli.py              # Command-line interface
+│   ├── pipeline.py         # High-level scraping orchestration
+│   ├── models.py           # Pydantic data models (PostLeaf, TopicLeaf, etc.)
+│   ├── parser_bbpress.py   # HTML parsing logic for bbPress forums
+│   ├── http_client.py      # Rate-limited HTTP client with ETag support
+│   ├── store_json.py       # JSON "leaf" persistence layer
+│   ├── index_sqlite.py     # SQLite FTS5 search index
+│   ├── normalize.py        # Content extraction (links, quotes, blocks)
+│   ├── export.py           # Export utilities (CSV, text, by-author, etc.)
+│   ├── coverage.py         # Coverage tracking for incremental scrapes
+│   ├── query_tracker.py    # Query provenance/lineage tracking
+│   ├── report.py           # Reporting utilities
+│   └── config.py           # Settings with env var overrides
+├── packages/                # TypeScript UI layer (pnpm workspace)
+│   ├── browser-app/        # React/Vite micro-frontend remote (port 3020) — UI via @ojfbot/frame-ui-components (npm ^1.0.1)
+│   ├── browser-automation/ # Visual regression testing via browser automation (screenshot capture, CI integration)
+│   ├── api/                # Express API over flat JSON + SQLite (port 3021) — includes GET /api/beads (ADR-0016)
+│   └── shared/             # @purefoy/shared — generated OpenAPI schema + API type contracts (incl. GoalManifest, ReviewProgress)
+├── app.py                   # Standalone Flask UI — dark-theme knowledge browser (port 5050)
+├── download_episodes.py     # Podcast RSS feed downloader (standalone)
+├── ingest_teamdeakins_downloads.py  # Episode ingest with metadata & transcripts
+├── scrape_forum.py          # Legacy forum scraper (standalone)
+├── library/                 # Output directory for scraped data
+│   └── forums/             # Forum JSON leafs organized by type
+│       ├── posts/          # Individual post JSON files
+│       ├── topics/         # Topic index files
+│       ├── forums/         # Forum metadata files
+│       └── _site/          # Sqlite index, HTTP state, coverage
+├── downloads/               # Podcast episode directories (348/348 episodes transcribed)
+│   └── S02E176__2026-01-14__guest-name__libsyn_abc123/
+│       ├── audio.mp3
+│       ├── metadata.json
+│       └── transcript/
+├── analysis/                # Export output (CSV, text dumps, stats)
+├── package.json             # pnpm workspace root
+└── pyproject.toml           # Python package metadata
+```
+
+## Podcast Episode Management
+
+```bash
+# Download MP3s from RSS feed
+.venv/bin/python3 download_episodes.py
+
+# Organize downloads into episode directories (dry-run)
+.venv/bin/python3 ingest_teamdeakins_downloads.py \
+  --rss "https://rss.libsyn.com/shows/265448/destinations/2018942.xml" \
+  --downloads "./downloads" \
+  --dry-run
+
+# Actually ingest with transcript fetching
+.venv/bin/python3 ingest_teamdeakins_downloads.py \
+  --rss "https://rss.libsyn.com/shows/265448/destinations/2018942.xml" \
+  --downloads "./downloads" \
+  --fill-transcripts \
+  --tapesearch
+```
+
+### Episode Ingest Design
+
+The `ingest_teamdeakins_downloads.py` script:
+- Parses RSS XML to extract episode metadata (title, date, duration, iTunes tags, description)
+- Matches downloaded MP3s to RSS entries via URL
+- Creates structured episode directories with naming: `S{season}E{episode}__{date}__{guest}__{source_id}/`
+- Generates `metadata.json` with comprehensive provenance
+- Creates `transcript/` scaffold with `sources.json` and `transcript.txt` (placeholder or fetched)
+- Supports Podcasting 2.0 `<podcast:transcript>` URLs and optional Tapesearch fallback
+
+## Testing Notes
+
+- **TypeScript layer**: CI runs TypeScript type-check, test, and codegen drift guard jobs (see `packages/`). Fleet-wide visual regression tests run via the `browser-automation` package (screenshots capture the composed shell, not standalone apps; waits for a Frame sentinel rather than `networkidle`).
+- **Python layer**: Manual testing via CLI commands and dry-run flags; no formal Python test suite yet
+- Incremental updates can be verified by running scrapes twice and checking for "Not modified" messages
diff --git a/documentation/roadmap/ROADMAP.md b/documentation/roadmap/ROADMAP.md
index f6d8c92..3bb00fa 100644
--- a/documentation/roadmap/ROADMAP.md
+++ b/documentation/roadmap/ROADMAP.md
@@ -370,15 +370,15 @@ python -m deakins_forums.cli tree team-deakins__bikes --format html -o bikes_tre
 
 ### Open Source Considerations
 
-#### 4.6 Public vs. Private Decision
+#### 4.6 Public vs. Private Decision — ✅ RESOLVED
 **Priority:** Critical (before git init)
 
-**Options:**
+**Options considered:**
 1. **Private Repository**: Keep scraped data private, respect forum ToS
-2. **Public Code, Private Data**: Open-source scrapers, private data repo
+2. **Public Code, Private Data**: Open-source scrapers, scraped data off-git
 3. **Fully Public**: Open everything (requires legal review)
 
-**Current:** Private repository assumed
+**Decision:** Option 2 — **public repository, code only**. Scrapers and docs are public; scraped forum content (`library/`), audio (`downloads/`), and exports (`analysis/`) are gitignored and never committed. Respects forum ToS/copyright while keeping the codebase open.
 
 ---