Reorganize docs by project and archive legacy context files

README.md

@@ -1,28 +1,22 @@
 # SearchXNG Project
 
-Unified workspace for:
+This repo is now organized by project area.
 
-- `searchxng/`: SearXNG settings, engine wiring, and instance branding context.
-- `api/`: SethSearch API context, indexing rules, and source integration details.
-- `context/`: operational runbooks and architecture docs.
+## Layout
 
-## Stack
+- `projects/searchxng/`: SearXNG integration, engine wiring, and service context.
+- `projects/sethsearch-api/`: SethSearch indexer/API behavior, source rules, and maintenance.
+- `projects/shared/`: cross-cutting architecture and operations runbooks.
+- `archive/legacy-context-2026-03-13/`: archived first-pass context docs.
 
-- SearXNG instance: `searxng.sethpc.xyz` / `search.sethpc.xyz` (CT 119 on `pve173`)
-- SethSearch API: `sethsearch.sethpc.xyz` (CT 620 on `pve173`)
-- Reverse proxy: Caddy CT 600 (`pve241`)
+## Live stack
 
-## Key behavior
+- SearXNG: `https://searxng.sethpc.xyz` and `https://search.sethpc.xyz` (CT 119, `pve173`)
+- SethSearch API: `https://sethsearch.sethpc.xyz` (CT 620, `pve173`)
+- Caddy: CT 600 (`pve241`)
 
-- `sethsearch` engine (general): searches homelab/site/content sources with strict FreshRSS and strict Emby matching.
-- `sethflix` engine (videos): Emby-only search for media discovery.
-- LibreTranslate integrated via `translate.sethpc.xyz`.
+## Quick commands
 
-## Quick Ops
-
-- Restart SethSearch API:
-  - `ssh pve173 "pct exec 620 -- systemctl restart sethsearch"`
-- Restart SearXNG:
-  - `ssh pve173 "pct exec 119 -- systemctl restart searxng"`
-- Validate Caddy:
-  - `ssh pve241 "pct exec 600 -- caddy validate --config /etc/caddy/Caddyfile"`
+- Restart SethSearch: `ssh pve173 "pct exec 620 -- systemctl restart sethsearch"`
+- Restart SearXNG: `ssh pve173 "pct exec 119 -- systemctl restart searxng"`
+- Validate Caddy: `ssh pve241 "pct exec 600 -- caddy validate --config /etc/caddy/Caddyfile"`

archive/README.md

@@ -0,0 +1,5 @@
+# Archive
+
+This folder stores superseded context docs.
+
+- `legacy-context-2026-03-13/`: initial ungrouped context/runbook files moved during project-based reorganization.

projects/searchxng/CONTEXT.md

@@ -0,0 +1,23 @@
+# Project: SearchXNG (SearXNG Layer)
+
+## Purpose
+
+Run and tune the SearXNG front-end search experience and route custom engines to SethSearch.
+
+## Deployment
+
+- Container: CT `119` on `pve173`
+- Service URL: `https://searxng.sethpc.xyz`, `https://search.sethpc.xyz`
+- Config file: `/etc/searxng/settings.yml`
+
+## Active custom engines
+
+- `sethsearch` (`shortcut: ss`, category: `general`, weight: `5.0`)
+- `sethflix` (`shortcut: sfx`, category: `videos`, weight: `5.0`)
+- `libretranslate` (`shortcut: lt`, category: `translate`)
+
+## Notes
+
+- Use HTTPS URLs for `json_engine` targets.
+- Keep engine names lowercase to avoid startup warnings.
+- This layer owns weighting and user-facing engine categories.

projects/searchxng/README.md

@@ -0,0 +1,10 @@
+# SearchXNG Project Docs
+
+- Runtime config: `/etc/searxng/settings.yml`
+- Primary task: expose and weight custom engines for homelab and media search.
+- Engine endpoint target: `https://sethsearch.sethpc.xyz/search`
+
+## Validation
+
+- Engine list check: `curl -s "http://192.168.0.177:8888/config"`
+- Restart service: `ssh pve173 "pct exec 119 -- systemctl restart searxng"`

projects/sethsearch-api/CONTEXT.md

@@ -0,0 +1,27 @@
+# Project: SethSearch API
+
+## Purpose
+
+Provide a unified `json_engine` backend for SearXNG with source-aware matching and weighting.
+
+## Deployment
+
+- Container: CT `620` on `pve173`
+- Service URL: `https://sethsearch.sethpc.xyz`
+- Service name: `sethsearch`
+- Code path: `/opt/sethsearch/sethsearch.py`
+- Config path: `/opt/sethsearch/config.json`
+
+## Source groups
+
+- `source=general`: sites, gitea, wikijs, wordpress, freshrss, emby (strict emby/freshrss)
+- `source=sethflix`: emby-only (liberal media matching)
+
+## Indexed sources
+
+- Caddy domains (`sites`)
+- Gitea public repos (`gitea`)
+- Wiki.js public pages/fallback (`wikijs`)
+- WordPress public pages/posts (`wordpress`)
+- Emby media catalog (`emby`)
+- FreshRSS feed items (`freshrss`)

projects/sethsearch-api/README.md

@@ -0,0 +1,17 @@
+# SethSearch API Docs
+
+- Health: `GET /health`
+- Search: `GET /search?q=<query>&source=<group|source>&limit=<n>`
+- Stats: `GET /stats`
+- Manual sync: `POST /sync`
+
+## Operations
+
+- Restart: `ssh pve173 "pct exec 620 -- systemctl restart sethsearch"`
+- Logs: `ssh pve173 "pct exec 620 -- journalctl -u sethsearch -n 100 --no-pager"`
+
+## Matching behavior
+
+- FreshRSS: strict + lower source weight.
+- Emby in general: strict.
+- Emby in sethflix: liberal and title-boosted.

projects/shared/ARCHITECTURE.md

@@ -0,0 +1,20 @@
+# Shared Architecture
+
+## Request flow
+
+1. User searches on SearXNG.
+2. Custom engines query SethSearch API (`json_engine`).
+3. SethSearch reads SQLite FTS5 index and applies source-specific scoring.
+4. SearXNG merges with other web engines and renders response.
+
+## Data model
+
+- DB: `/opt/sethsearch/articles.db`
+- Main table: `documents`
+- FTS table: `documents_fts`
+
+## Reliability pattern
+
+- Source sync jobs run independently.
+- Failures in one source do not block other source updates.
+- systemd restart policy keeps service resilient.

projects/shared/OPERATIONS.md

@@ -0,0 +1,16 @@
+# Shared Operations
+
+## SearXNG checks
+
+- Status: `ssh pve173 "pct exec 119 -- systemctl status searxng --no-pager"`
+- Logs: `ssh pve173 "pct exec 119 -- journalctl -u searxng -n 100 --no-pager"`
+
+## SethSearch checks
+
+- Status: `ssh pve173 "pct exec 620 -- systemctl status sethsearch --no-pager"`
+- Logs: `ssh pve173 "pct exec 620 -- journalctl -u sethsearch -n 100 --no-pager"`
+- Stats: `curl -s "https://sethsearch.sethpc.xyz/stats"`
+
+## Caddy checks
+
+- Validate: `ssh pve241 "pct exec 600 -- caddy validate --config /etc/caddy/Caddyfile"`