io.github.cyanheads/seerr-mcp-server

@cyanheads/seerr-mcp-server

Search Jellyseerr/Overseerr, check availability, and create guarded media requests via MCP. STDIO or Streamable HTTP.

6 Tools • 1 Resource

Overview

Jellyseerr and Overseerr media-request workflow: search TMDB-backed titles, confirm the exact match, check availability and request state, and create a guarded request that Radarr/Sonarr act on. Jellyseerr owns permissions, quotas, routing, and status — this server never calls Radarr/Sonarr directly. Runs as a stdio process or a local Streamable HTTP server.

Tools

Resources

All request data is also reachable via tools — request enumeration is the job of seerr_list_requests (the tool-only access path).

Capability reference

seerr_search_media tool

• Search titles with mediaType: movie | tv | all; people are excluded. page is 1–1000; limit is 1–20, default 10.
• Returns TMDB matches and decoded status / status4k for tracked titles. Empty results return [] with a guidance notice.
• Optional ISO 639-1 language selects localized titles and overviews.

seerr_get_media tool

• Availability plus any existing open request for the title (avoids duplicate requests)
• TV: omit seasonNumber for a per-season summary, or pass one to fetch that season's episode list (season 0 is Specials)
• A TMDB ID that doesn't resolve surfaces as a typed media_not_found with a search-recovery hint (Jellyseerr's raw HTTP 500 is classified in the service layer)

seerr_list_requests tool

• Filter by lifecycle filter, mediaType, and requestedById; sort by added or modified, ascending or descending. take is 1–100, default 20; skip selects the offset.
• Returns decoded statuses, requester { id, displayName }, and an enrichment trailer with applied filters.
• includeTitles: true joins media titles once per distinct title (default off); unresolved rows retain their data and are disclosed in the notice.

seerr_request_media tool

• Guarded write. mode: preview (default) returns the resolved title and exact POST /request payload without writing. mode: request asks for confirmation bound to the same caller, title and payload; declined, cancelled or invalid answers return request_cancelled.
• TV requests require seasons: "all" or a list such as [1, 2]; season 0 (Specials) requires special episodes to be enabled. Instance capability checks run before any POST; existing requests appear in output and duplicate rejections return duplicate_request.
• Optional routing overrides: serverId, profileId, rootFolder, languageProfileId. Omit them to use Jellyseerr's defaults.

seerr_request_status tool

• Accepts requestId from seerr_request_media's created.requestId or seerr_list_requests. A missing request returns request_not_found.
• Returns decoded requestStatus, mediaStatus / mediaStatus4k, requester { id, displayName }, routing (serverId, profileName, is4k), and stateGuidance.
• Joins title on each call; it is absent when no tmdbId exists or the lookup fails.

seerr_service_options tool

• Returns instance version, media server, capability flags (movie4kEnabled, series4kEnabled, partialRequestsEnabled, specialEpisodesEnabled), and service IDs, defaults and quality profiles.
• Failed service legs retain the available results with a notice.
• includePaths: true exposes root-folder paths and free space; both are omitted by default.

seerr://request/{requestId} resource

• Accepts requestId from seerr_request_media or seerr_list_requests; a missing request returns request_not_found.
• Mirrors seerr_request_status, including redaction and a title join. title is absent when no tmdbId exists or the lookup fails; resources take no per-read options.

Features

Built on @cyanheads/mcp-ts-core: stdio and Streamable HTTP transports, pluggable auth (none / jwt / oauth), swappable storage (in-memory, filesystem, Supabase, Cloudflare KV/R2/D1), structured logging with optional OpenTelemetry tracing.

HTTP requires stateful sessions for 2025-era confirmation rounds; explicit MCP_SESSION_MODE=stateless fails startup. Consent records expire after ten minutes and changed or replayed records ask again. Redemption excludes concurrent retries in one process; request creation makes one POST attempt without automatic retries. Cross-instance read-and-delete is not atomic: use shared strongly consistent filesystem, Supabase or Cloudflare D1 storage for routed retries, and account for duplicate-submission races; Cloudflare KV is unsuitable for consent records.

Seerr-specific:

• Read + guarded-request only — admin-scope endpoints (approve/decline, retry, edit/delete, media/file deletion, user/settings/sync) are excluded by design, not by API limitation
• PII/infra redaction centralized in one normalizer choke point — requesters project to { id, displayName }; operator email, Plex/Jellyfin tokens, and serviceUrl never reach output; filesystem paths are opt-in via includePaths
• Status decoding centralized in one helper — request and media statuses (including the separate 4K availability) decode to { raw, label } everywhere, forward-compatible with new Jellyseerr status codes
• Capability validation against cached instance settings catches most bad requests before they reach the API
• A short-TTL settings cache avoids a round-trip on every preview

Agent-friendly output:

• Provenance and disclosure — searches echo the effective query; capped lists disclose truncation; a degraded service leg surfaces a notice instead of silently dropping data
• Typed, actionable errors — media_not_found, request_not_found, seasons_required, four_k_not_enabled, duplicate_request, and more carry a recovery hint so callers can branch and retry without parsing prose

Getting started

Add the following to your MCP client configuration file, pointing SEERR_BASE_URL at your own Jellyseerr or Overseerr instance and supplying its API key.

{
  "mcpServers": {
    "seerr-mcp-server": {
      "type": "stdio",
      "command": "bunx",
      "args": ["@cyanheads/seerr-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info",
        "SEERR_BASE_URL": "http://localhost:5055",
        "SEERR_API_KEY": "your-api-key"
      }
    }
  }
}

Or with npx (no Bun required):

{
  "mcpServers": {
    "seerr-mcp-server": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@cyanheads/seerr-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info",
        "SEERR_BASE_URL": "http://localhost:5055",
        "SEERR_API_KEY": "your-api-key"
      }
    }
  }
}

For Streamable HTTP, set the transport and start the server:

MCP_TRANSPORT_TYPE=http MCP_HTTP_PORT=3010 SEERR_BASE_URL=http://localhost:5055 SEERR_API_KEY=your-api-key bun run start:http
# Server listens at http://localhost:3010/mcp

Prerequisites

• Bun v1.4.0 or higher (or Node.js v24+).
• A running Jellyseerr or Overseerr instance, and its API key (Settings → General → API Key).

Installation

1. Clone the repository:

git clone https://github.com/cyanheads/seerr-mcp-server.git

1. Navigate into the directory:

cd seerr-mcp-server

1. Install dependencies:

bun install

1. Configure environment:

cp .env.example .env
# edit .env — set SEERR_BASE_URL and SEERR_API_KEY

Configuration

All configuration is validated at startup via Zod schemas in src/config/server-config.ts. Key environment variables:

See .env.example for the full list of optional overrides.

Running the server

Local development

• Build and run:

  # One-time build
  bun run rebuild
  
  # Run the built server
  bun run start:stdio
  # or
  bun run start:http

• Run checks and tests:

  bun run devcheck   # Lint, format, typecheck, security, changelog sync
  bun run test       # Vitest test suite
  bun run lint:mcp   # Validate MCP definitions against spec

Docker

docker build -t seerr-mcp-server .
docker run --rm \
  -e SEERR_BASE_URL=http://host.docker.internal:5055 \
  -e SEERR_API_KEY=your-api-key \
  -p 3010:3010 \
  seerr-mcp-server

The Dockerfile defaults to HTTP transport, stateful session mode, and logs to /var/log/seerr-mcp-server. OpenTelemetry peer dependencies are installed by default — build with --build-arg OTEL_ENABLED=false to omit them.

Project structure

Development guide

See CLAUDE.md/AGENTS.md for development guidelines and architectural rules. The short version:

• Handlers throw, framework catches — no try/catch in tool logic
• Use ctx.log for request-scoped logging, ctx.state for tenant-scoped storage
• Register new tools and resources via the barrels in src/mcp-server/*/definitions/index.ts
• Wrap the Jellyseerr API: validate raw → normalize and redact to a domain type → return the output schema; never fabricate missing fields, and never let operator PII or paths reach output

Contributing

Issues are welcome. Run checks and tests before submitting:

bun run devcheck
bun run test

License

Apache-2.0 — see LICENSE for details.