FlixBridge
FlixBridge is a stdio MCP server for operating Sonarr, Radarr, and SABnzbd (with optional Plex issue triage) through MCP tools for monitoring, searching, adding, importing, removing, and diagnosing media.
List configured services and downloaders.
Check system status, server metrics, and unified download status.
View queues, history, root folders, quality profiles, and import issues.
Search for series/movies and add new media to libraries.
Force grab/retry queued items.
Preview and execute manual imports with dry-run and confirmation token.
Preview and remove queue/library items with dry-run, confirmation token, and optional blocklist/delete files.
Run read-only queue diagnostics by default, or opt into auto-fix per service or across all services.
Supports multiple ARR instances via environment variables; no web dashboard or background polling.
Mutations require operator authority; verify affected records after every change.
Provides comprehensive movie management capabilities including adding new movies, managing quality profiles, monitoring download queues, running diagnostics, and accessing system health information
Enables TV show management through tools for adding series, configuring quality profiles, monitoring downloads, managing libraries, and performing automated diagnostics and troubleshooting
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@FlixBridge check my download queue status"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
FlixBridge
A stdio MCP server for operating Sonarr, Radarr, and SABnzbd, with optional Plex reported-issue triage. Supports multiple ARR instances through environment variables. Requires Node.js 20 or later and at least one configured Sonarr or Radarr instance.
Connect an MCP client
Use this server entry in a client that supports mcpServers (for example Claude Desktop). Replace the placeholders with your service URL and API key; keep credentials private.
{
"mcpServers": {
"flixbridge": {
"command": "npx",
"args": ["-y", "@thesammykins/flixbridge"],
"env": {
"SONARR_URL": "http://localhost:8989",
"SONARR_API_KEY": "replace-with-sonarr-api-key"
}
}
}
}This runs the npm release, not a local checkout. For reproducible deployments, select and pin a published package version. For source changes, build the checkout and configure the client to run node /absolute/path/to/FlixBridge/dist/index.js; see installation.
Related MCP server: Trakt
Operations and safety
Call list_services first and use the returned names. Tools cover queue/status monitoring, search/add, quality profiles, history, imports, removal, diagnostics, and Plex reports. There is no web dashboard or background polling service.
Diagnostics default to read-only (
autoFix:false); keep it explicit for inspection. Opting intoautoFix:truerequires authorization.manual_importandremove_contentrequire a dry run and a returned confirmation token before execution.Add, retry/search, profile changes, and Plex comments also mutate state. Require operator authority even when a tool has no token gate.
Read back the exact affected records after every mutation. An accepted command or a report marked “fixed” does not establish successful playback.
Optional configuration: RADARR_URL / RADARR_API_KEY, SABNZBD_URL / SABNZBD_API_KEY, and PLEX_BASE_URL / PLEX_TOKEN. Multi-instance variables use SONARR_<SLUG>_URL / SONARR_<SLUG>_API_KEY (also Radarr and SABnzbd). Configuration is environment-only; .env files are not loaded automatically.
Documentation
Source status and compatibility
This documentation describes the 1.0.0 source and tagged release. Its 20 tools retain the existing 19 names/required inputs and add read-only library_statistics for Sonarr API v5 (no legacy library scan). GitHub release assets and npm registry versions are separate; an unpinned npm command may select an older release. See tagged-release installation.
Sonarr API policy is per instance: omission preserves v3 without discovery; explicit 3 or 5 discovers and verifies that family; auto uses v5 only for reviewed workflows and retains v3 for unported operations when advertised. Strict v5 rejects those operations before writes. A Sonarr application version is not its HTTP API family. See configuration and route limits.
The SDK serves current MCP 2026-07-28 and legacy stdio clients, with bounded dispatch and request cancellation. Cancellation is not rollback: accepted upstream work may continue. Pending imports remain in legacy failed counts with additive pending evidence; do not resubmit blindly. See transport and validation, migration and rollback, and security boundaries.
CI verifies only; publishing an exact reviewed artifact/version requires separate final maintainer approval.
Development
git clone https://github.com/thesammykins/FlixBridge.git
cd FlixBridge
npm ci --include=dev --ignore-scripts
npm run build
npm run check
npm testRead AGENTS.md before contributing. Unit tests use mocks; MCP tests exercise local stdio/client/HTTP boundaries without a live media stack. Live scripts are separate. npm run smoke uses read-only diagnostics by default in this source, but contacts configured services and enables debug logs; inspect it and obtain authority before live use.
Support
If Flixbridge is useful to you, you can support its development.
Available Tools
16 toolsadd_newC
Add new media to library
Inputs: title (required), service (required), foreignId (required), monitored, +2 more.
all_services_diagnosticsB
Analyze and auto-fix stuck queue items across all services
Inputs: autoFix.
download_statusC
Get unified download status across arr services and downloaders
Inputs: services, downloader, includeDownloader.
history_detailC
Get download/import history details
Inputs: service (required), page, since, pageSize.
import_issuesC
Check for import issues and stuck downloads
Inputs: service (required).
list_servicesA
List all configured services and downloaders. Call this first to see available services.
Inputs: no named parameters.
manual_importC
Preview and execute high-confidence ARR manual imports for queue items
Inputs: ids (required), service (required), dryRun, confirmationToken.
quality_profilesC
List available quality profiles with recommendations
Inputs: service (required).
queue_diagnosticsC
Analyze and auto-fix stuck queue items
Inputs: service (required), autoFix.
queue_grabC
Force grab/retry download of queued items
Inputs: ids (required), service (required).
queue_listC
List items in download queue with status and progress
Inputs: service (required), page, pageSize.
remove_contentC
Preview and remove queue or library items with confirmation
Inputs: ids (required), dryRun, target, service, +10 more.
root_foldersC
List configured root folders and storage information
Inputs: service (required).
searchC
Search for media (series/movies) to add
Inputs: query (required), service (required), limit.
server_metricsC
Get server performance metrics and health status
Inputs: service, detailed.
system_statusC
Get system status and health information
Inputs: service (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 16 tools
Several tools overlap heavily, especially all_services_diagnostics, queue_diagnostics, and import_issues, which all deal with stuck queue items and import failures. system_status, server_metrics, and download_status also blur the line between health/performance and download status.
Names are a mix of noun-first, verb-first, and adjective-noun forms without a consistent pattern. Examples like queue_list, list_services, and add_new show inconsistent conventions; queue_grab reverses the expected verb_noun order.
16 tools is on the higher end for the apparent scope, and several diagnostic/status tools could be consolidated. The count is not absurd, but it feels heavier than necessary due to redundancy.
Core media management workflows are covered: search, add, queue management, imports, history, removal, and configuration listings. A notable gap is direct library listing/management beyond content removal, though the surface covers the main ARR-type workflows.
Maintenance
Related MCP Connectors
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
One MCP server for the services your AI needs: memory, files, Microsoft 365, kanban and more.
The Remote MCP server acts as a standardized bridge between LLM applications (like Claude, ChatGPT, and Cursor) and external services, enabling AI agents to access external tools and resources. Its primary capability is providing a centralized search tool to discover other MCP servers and their respective tools. Unlike local implementations, it runs remotely with OAuth authentication and permission controls for security.
Nifty's MCP server — exposes tasks, projects, messages, and files as tools for AI agents.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceA comprehensive Model Context Protocol server that bridges LLMs with self-hosted media services, enabling natural language control of TV shows, movies, downloads, and notifications while maintaining traditional API access.17AGPL 3.0
- AlicenseNot gradedqualityBmaintenanceA Model Context Protocol (MCP) server that creates a bridge between AI language models and the Trakt.tv API, allowing LLMs to access real-time entertainment data and personal Trakt viewing history.51MIT
- AlicenseBqualityFmaintenanceA bridge server that connects Agent Communication Protocol (ACP) agents with Model Context Protocol (MCP) clients, enabling seamless integration between ACP-based AI agents and MCP-compatible tools like Claude Desktop.1665 PyPI24MIT
- AlicenseBqualityAmaintenanceA Model Context Protocol server that enables AI assistants to query and manage Plex Media Server content through natural language, providing library access, viewing statistics, and media management capabilities.54253 npm58JavaScriptMIT