Seerr MCP Server
# Seerr MCP Server
[](https://opensource.org/licenses/MIT)
[](https://github.com/jhomen368/overseerr-mcp/pkgs/container/overseerr-mcp)
[](https://github.com/jhomen368/overseerr-mcp)
[](https://www.paypal.com/donate?hosted_button_id=PBRD7FXKSKAD2)
> **A [Model Context Protocol](https://modelcontextprotocol.io) (MCP) server for Overseerr and Seerr (the unified successor) that enables AI assistants to search, request, and manage media through the Model Context Protocol.**
## π― Key Features
- **π 99% fewer API calls** for batch operations (150-300 β 1)
- **β‘ 88% token reduction** with compact response formats
- **π― Batch Dedupe Mode** - Check 50-100 titles in one operation
- **π Smart Caching** - 70-85% API call reduction
- **π‘οΈ Safety Features** - Multi-season confirmation, validation
- **π¦ 6 Tools** - Search, request, and manage media | Discover Radarr/Sonarr server configurations
## π Security
- **π€ Automated Security Scanning**
- Dependabot for dependency updates (weekly)
- CodeQL for code vulnerability analysis (PR + weekly)
- Trivy for Docker image scanning (CI only - blocks PRs if vulnerabilities found)
- CI validates everything during PR review, CD trusts CI and publishes
- **π³ Hardened Docker Images**
- Non-root user (mcpuser)
- Multi-stage builds
- Minimal Alpine base
- dumb-init process management
- **β
Input Validation**
- URL and API key format validation
- Fails fast with clear error messages
## π οΈ Available Tools
| Tool | Purpose | Key Features |
|------|---------|--------------|
| **search_media** | Search & dedupe | Single/batch search, dedupe mode for 50-100 titles, franchise awareness |
| **request_media** | Request movies/TV | Batch requests, season validation, multi-season confirmation, dry-run mode |
| **manage_media_requests** | Manage requests | List/approve/decline/delete, filtering, summary statistics |
| **get_media_details** | Get media info | Batch lookup, flexible detail levels (basic/standard/full) |
| **get_services** | List Radarr/Sonarr servers | Discover server IDs, active defaults, 4K status |
| **get_service_details** | Get server config | Quality profiles, root folders, tags per server |
## π Prerequisites
- **Node.js** 18.0 or higher
- **Seerr or Overseerr instance** (self-hosted or managed)
- **Seerr/Overseerr API key** (Settings β General in your instance)
## π Quick Start
### Option 1: NPM (Recommended)
```bash
npm install -g @jhomen368/overseerr-mcp
```
**Configure with Claude Desktop:**
Add to your configuration file:
- **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
```json
{
"mcpServers": {
"seerr": {
"command": "npx",
"args": ["-y", "@jhomen368/overseerr-mcp"],
"env": {
"SEERR_URL": "https://seerr.example.com",
"SEERR_API_KEY": "your-api-key-here"
}
}
}
}
```
> **Legacy Overseerr Users:** If you're still using Overseerr (not Seerr), you can continue using the legacy variables:
> ```json
> {
> "env": {
> "OVERSEERR_URL": "https://overseerr.example.com",
> "OVERSEERR_API_KEY": "your-api-key-here"
> }
> }
> ```
> *Both `OVERSEERR_*` and `SEERR_*` variables are supported for backward compatibility. Legacy variables will be removed in v3.0.0.*
### Option 2: Docker (Remote Access)
```bash
docker run -d \
--name seerr-mcp \
-p 8085:8085 \
-e SEERR_URL=https://your-seerr-instance.com \
-e SEERR_API_KEY=your-api-key-here \
ghcr.io/jhomen368/overseerr-mcp:latest
```
**Docker Compose:**
```yaml
services:
seerr-mcp:
image: ghcr.io/jhomen368/overseerr-mcp:latest
container_name: seerr-mcp
ports:
- "8085:8085"
environment:
- SEERR_URL=https://your-seerr-instance.com
- SEERR_API_KEY=your-api-key-here
restart: unless-stopped
```
**Test the server:**
```bash
curl http://localhost:8085/health
```
**Connect MCP clients:**
- **Transport**: Streamable HTTP
- **URL**: `http://localhost:8085/mcp`
### Option 3: From Source
```bash
git clone https://github.com/jhomen368/overseerr-mcp.git
cd overseerr-mcp
npm install
npm run build
node build/index.js
```
## π‘ Usage Examples
### Batch Dedupe Workflow (Perfect for Anime Seasons)
```typescript
// Check 50-100 titles in ONE API call
search_media({
dedupeMode: true,
titles: [
"Frieren: Beyond Journey's End",
"My Hero Academia Season 7",
"Demon Slayer Season 4",
// ... 47 more titles
],
autoNormalize: true // Strips "Season N", "Part N", etc.
})
```
**Response:**
```json
{
"summary": {
"total": 50,
"pass": 35,
"blocked": 15,
"passRate": "70%"
},
"results": [
{ "title": "Frieren", "status": "pass", "id": 209867 },
{ "title": "My Hero Academia S7", "status": "pass", "franchiseInfo": "S1-S6 in library" },
{ "title": "Demon Slayer S4", "status": "blocked", "reason": "Already requested" }
]
}
```
### Request Media with Validation
```typescript
// Single movie request
request_media({
mediaType: "movie",
mediaId: 438631
})
// TV show with specific seasons
request_media({
mediaType: "tv",
mediaId: 82856,
seasons: [1, 2]
})
// All seasons (excludes season 0 by default)
request_media({
mediaType: "tv",
mediaId: 82856,
seasons: "all"
})
```
### Manage Requests
```typescript
// List with filters
manage_media_requests({
action: "list",
filter: "pending",
take: 20
})
// Batch approve
manage_media_requests({
action: "approve",
requestIds: [123, 124, 125]
})
// Get summary statistics
manage_media_requests({
action: "list",
summary: true
})
```
### Service Discovery
```typescript
// List all configured servers (Radarr + Sonarr)
get_services({})
// List only Radarr servers
get_services({ serviceType: "radarr" })
// Get quality profiles, root folders, and tags for a server
get_service_details({
serviceType: "radarr",
serverId: 0
})
// Use discovered values when requesting media
request_media({
mediaType: "movie",
mediaId: 438631,
serverId: 0,
profileId: 13,
rootFolder: "/data/media/movies"
})
```
### Natural Language Examples
Simply ask your AI assistant:
- "Search for Inception in Seerr"
- "Check if these 50 anime titles have been requested"
- "Request Breaking Bad all seasons"
- "Show me all pending media requests"
- "Approve request ID 123"
- "Get details for TMDB ID 550"
- "What Radarr servers are configured?"
- "Show me the quality profiles for my Sonarr server"
## βοΈ Configuration
### Environment Variables
**Required:**
- `SEERR_URL` - Your Seerr/Overseerr instance URL
- `SEERR_API_KEY` - API key from Settings β General
**Legacy (deprecated, will be removed in v3.0.0):**
- `OVERSEERR_URL` - Use `SEERR_URL` instead
- `OVERSEERR_API_KEY` - Use `SEERR_API_KEY` instead
**Optional (with defaults):**
```bash
CACHE_ENABLED=true # Enable caching
CACHE_SEARCH_TTL=300000 # Search cache: 5 min
CACHE_MEDIA_TTL=1800000 # Media cache: 30 min
CACHE_REQUESTS_TTL=60000 # Request cache: 1 min
CACHE_MAX_SIZE=1000 # Max cache entries
CACHE_SERVICES_TTL=600000 # Services cache: 10 min
CACHE_SERVICEDETAILS_TTL=600000 # Service details cache: 10 min
REQUIRE_MULTI_SEASON_CONFIRM=true # Confirm >24 episodes
HTTP_MODE=false # Enable HTTP transport
PORT=8085 # HTTP server port
```
## π Documentation
- **[CHANGELOG.md](CHANGELOG.md)** - Version history and release notes
- **[CONTRIBUTING.md](CONTRIBUTING.md)** - Contribution guidelines
- **[Overseerr API Docs](https://api-docs.overseerr.dev/)** - Official API reference
## π§ Troubleshooting
### Connection Issues
- Verify Seerr/Overseerr URL is accessible
- Check API key validity (Settings β General)
- Review firewall rules for remote access
### Docker Issues
```bash
# Check logs
docker logs seerr-mcp
# Verify health
curl http://localhost:8085/health
# Restart container
docker restart seerr-mcp
```
### Build Issues
```bash
# Ensure Node.js 18+
node --version
# Clean rebuild
rm -rf node_modules build
npm install
npm run build
```
## π€ Contributing
Contributions welcome! Please see [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
## π License
MIT License - see [LICENSE](LICENSE) for details
## π Acknowledgments
- [Seerr](https://github.com/seerr) - Next-generation media request and discovery tool
- [Overseerr](https://overseerr.dev/) - Original media request tool for Plex
- [Model Context Protocol](https://modelcontextprotocol.io) - Open protocol for AI integrations
- [Anthropic](https://www.anthropic.com/) - Creators of the MCP standard
---
**Support this project:** [](https://www.paypal.com/donate?hosted_button_id=PBRD7FXKSKAD2)
TDQS
Scored across 6 tools
Each tool maps to a distinct action (list services, search, request, manage requests, details, service config), but the pairs get_services/get_service_details and search_media/get_media_details have overlapping-sounding names that require reading descriptions to separate. The descriptions do clarify the boundary (list vs. config detail; quick status vs. full detail), so misselection risk is low but non-zero.
All names use snake_case with a verb_noun pattern (get_services, search_media, request_media, manage_media_requests), which is highly predictable. Minor inconsistency in granularity between get_services (plural list) and get_service_details (singular detail), but the convention holds.
Six tools is a tight, well-scoped surface for a Seerr/Radarr/Sonarr request-proxy server, with no redundant or filler tools. Each tool covers a distinct stage of the discoverβrequestβmanageβinspect workflow.
The core lifecycle is covered: service discovery, media search with actionable statuses, request creation with confirmation rules, request management (get/list/approve/decline/delete), and detail inspection. Gaps exist around library-side operations (deleting media from Radarr/Sonarr) and user/quota or settings management, but agents can work around these.