## Approach
- Read existing files before writing. Don't re-read unless changed.
- Thorough in reasoning, concise in output.
- Skip files over 100KB unless required.
- No sycophantic openers or closing fluff.
- No emojis or em-dashes.
- Do not guess APIs, versions, flags, commit SHAs, or package names. Verify by reading code or docs before asserting.

## Global
- Utilise Docker & Docker-Compose pour la base de données (MariaDB)
- On développe tout en NodeJS notamment avec : NextJS & Prisma pour la partie backend et React pour la partie frontend
- Utilise Tailwind pour le CSS
- Si tu dois utiliser des Workers, gère le avec Passenger (pm2)
- L'application est toujours multilangue (EN, FR, ES, IT — voir `messages/`)

## Application : Ombrora-YTDLP

- Interface web publique permettant de soumettre des URLs de videos a telecharger via yt-dlp, avec pages SEO par plateforme (`/[locale]/[slug]`, ex. `/youtube-downloader`, definies dans `src/lib/downloader-platforms.ts`) et une page `supported-sites` listant tous les sites geres par yt-dlp.
- Avant soumission, l'URL est analysee en temps reel via un probe yt-dlp (`POST /api/probe`, `src/lib/ytdlp-probe.ts`, `yt-dlp -J`) : les options presentees a l'utilisateur (qualite, sous-titres disponibles, decoupe, qualite audio MP3) sont derivees de cette analyse et non d'une liste statique ; la soumission est bloquee tant que le probe n'a pas reussi.
- Les telechargements sont geres via une file d'attente (queue) executee par un worker Node.js (`worker/`, lance via `tsx`) supervise par pm2, avec support du multithreading (concurrence configurable via `WORKER_CONCURRENCY`).
- Chaque telechargement est stocke en base de donnees (MariaDB via Prisma). **Aucune entree n'est jamais supprimee** : la DB conserve l'historique complet de tous les telechargements (statuts, erreurs, metadata).
- `yt-dlp` est le zipapp Python officiel (`bin/yt-dlp`), pas le binaire standalone PyInstaller : ce dernier s'auto-extrait dans `/tmp` et echoue avec "failed to map segment from shared object" quand `/tmp` est monte `noexec` (cas d'o2switch). Le zipapp est toujours invoque explicitement via `PYTHON_BIN` (`src/lib/ytdlp.ts`) plutot que via son shebang, car le `python3` systeme d'o2switch (3.6) est trop ancien pour yt-dlp (3.10+ requis) — sur o2switch, `PYTHON_BIN=/opt/alt/python311/bin/python3`.

## Deployment (o2switch)

- o2switch is shared hosting: no compiler toolchain, no root access. Any dependency with a native/binary component must ship as a precompiled binary — it cannot be built from source on the server. This is why `yt-dlp`/`ffmpeg`/`ffprobe` ship as precompiled binaries in `bin/` (or `BIN_DIR`) rather than as npm deps.
- Before adding any new dependency with native bindings, confirm it publishes prebuilt binaries for o2switch's platform/arch — otherwise it will fail to install or run there.
- Single Next.js app with a single `package.json`/`node_modules` at the repo root — there is no separate frontend build or subfolder to keep in sync. Deployed via cPanel "Setup Node.js App" (Passenger) for the web app; the worker is a separate process that Passenger does not supervise, kept alive by pm2 instead (`ecosystem.config.cjs`, `npm run worker:pm2:*`). Full deployment steps (including `npm run deploy` / `scripts/deploy.sh`) are documented in `README.md`.

<!-- BEGIN:nextjs-agent-rules -->

# This is NOT the Next.js you know

This version has breaking changes — APIs, conventions, and file structure may all differ from your training data. Read the relevant guide in `node_modules/next/dist/docs/` (resolved from this file's directory; in monorepos the `next` package may not be visible from the repo root) before writing any code. Heed deprecation notices.

This block is written and re-added by `next dev` — verify at `node_modules/next/dist/server/lib/generate-agent-files.js`. Removing it from a diff only re-creates the uncommitted change; committing it with your work keeps the tree clean.

<!-- END:nextjs-agent-rules -->

<!-- code-review-graph MCP tools -->
## MCP Tools: code-review-graph

**IMPORTANT: This project has a knowledge graph. ALWAYS use the
code-review-graph MCP tools BEFORE using Grep/Glob/Read to explore
the codebase.** The graph is faster, cheaper (fewer tokens), and gives
you structural context (callers, dependents, test coverage) that file
scanning cannot.

### When to use graph tools FIRST

- **Exploring code**: `semantic_search_nodes` or `query_graph` instead of Grep
- **Understanding impact**: `get_impact_radius` instead of manually tracing imports
- **Code review**: `detect_changes` + `get_review_context` instead of reading entire files
- **Finding relationships**: `query_graph` with callers_of/callees_of/imports_of/tests_for
- **Architecture questions**: `get_architecture_overview` + `list_communities`

Fall back to Grep/Glob/Read **only** when the graph doesn't cover what you need.

### Key Tools

| Tool | Use when |
|------|----------|
| `detect_changes` | Reviewing code changes — gives risk-scored analysis |
| `get_review_context` | Need source snippets for review — token-efficient |
| `get_impact_radius` | Understanding blast radius of a change |
| `get_affected_flows` | Finding which execution paths are impacted |
| `query_graph` | Tracing callers, callees, imports, tests, dependencies |
| `semantic_search_nodes` | Finding functions/classes by name or keyword |
| `get_architecture_overview` | Understanding high-level codebase structure |
| `refactor_tool` | Planning renames, finding dead code |

### Workflow

1. The graph auto-updates on file changes (via hooks).
2. Use `detect_changes` for code review.
3. Use `get_affected_flows` to understand impact.
4. Use `query_graph` pattern="tests_for" to check coverage.