A monitoring and statistics dashboard for Silo and Plex in one place: live streams, server health, full playback history and Tautulli-style stats per user, with Silo and Plex accounts linked to the same person. Access is behind your Silo administrator sign-in, two-step included. Built mobile-first, it runs comfortably on a phone.
Siloscope started as a fork of safetyp1/siloscope and has since grown a history database, a stats section, account linking, Silo sign-in and themes. Thanks to the original author for the foundation.
| Now Playing | Silo and Plex streams together: poster, user, client, quality, direct play or transcode (including hardware), bitrate and progress. Terminate one or several streams. |
| History | Every play from both servers, kept indefinitely in SQLite: measured watch and pause time, progress, client, platform, quality, LAN/WAN and IP. Imports Silo's full history and Plex's watch history on first start. |
| Stats | Plays and watch time over time, concurrent streams and the peak, plays by weekday and hour, stream quality, top platforms, most watched and most popular movies and shows, most active users, for a week, month, year or all time. |
| Users | Everyone who watches, on either server, with a page per person: their accounts, charts, top titles, devices and IP addresses. |
| Linked accounts | Silo and Plex accounts of the same person are matched automatically (by Plex sign-in or email) and can be linked, unlinked or renamed by hand, so one person is one row in the stats. |
| Server health | CPU, memory, bandwidth and top processes on the Silo host; disk free space with the size of each Silo library; Silo remote stream nodes. |
| Themes | Rose, Midnight, Light, High contrast, Console, Matrix and Synthwave, any accent colour, and per-device layout: reorder or hide tabs and panels. |
| Sign-in | Sign in with your Silo administrator account, including Silo's two-step sign-in. Other Silo accounts are refused, and sessions end if an account loses admin. |
| Optional extras | Push alerts, InfiniDysk throughput, Sonarr/Radarr file replacement and Silo transcoder settings. |
All screenshots use a demo server with made-up users and titles.
| System | Bandwidth and memory | Now Playing |
|---|---|---|
![]() |
![]() |
![]() |
| History | Stats | Charts |
|---|---|---|
![]() |
![]() |
![]() |
| Top titles | Users | A person |
|---|---|---|
![]() |
![]() |
![]() |
| Stream nodes | Appearance | Sign-in |
|---|---|---|
![]() |
![]() |
![]() |
Themes
| Silo | Plex |
|---|---|
![]() |
|
| Required: health, playback, history, nodes and transcoder settings | Optional: playback, history and stats |
Logos identify compatible services; no affiliation or endorsement is implied. See logo sources.
- Silo with the v2 administrator API. Older v1-only servers are not supported.
- A Silo API key with Full access, owned by an enabled administrator.
- Optional, recommended: a second Silo key with only the
admin:monitor:readscope, for live session events (see Silo API keys). - Optional: a Plex server-owner token, and outbound HTTPS to plex.tv for account linking.
- Docker on AMD64 or ARM64, or any Linux machine for the release binary.
Sign-in uses your Silo administrator account (see Sign-in). Anyone signed in can stop streams, change Silo settings, restart Silo, and delete media files when replacements are enabled. Serve it over HTTPS if it is reachable from outside your network.
Run Siloscope on the machine that runs Silo. It can then reach Silo over Silo's own Docker network, and its CPU, process and bandwidth panels measure the Silo host itself.
-
Clone the repository on the Silo host and create the environment file:
git clone https://github.com/neilmc78/siloscope /opt/siloscope && cd /opt/siloscope cp .env.example .env && chmod 600 .env
-
Find Silo's container, its network and the host's network interface:
docker ps --format '{.Names} {.Networks} {.Ports}' ip route show default -
Edit
.env:Setting Value SILO_URLSilo's container name and internal port, for example http://silo-silo-1:8080SILO_API_KEY/SILO_MONITOR_API_KEYYour two Silo keys PLEX_URL/PLEX_TOKENFor example http://192.168.1.50:32400and your server-owner tokenHOST_NETWORK_INTERFACEThe name after dev, for exampleeth0MONITOR_BIND/MONITOR_PORT0.0.0.0(or the host's LAN or Tailscale IP) and a free port, to open it from other devices -
Attach Siloscope to Silo's network: create
compose.override.yamlwith the network name from step 2:services: silo-monitor: networks: [silo] networks: silo: external: true name: silo_default
-
Start it with host bandwidth and process monitoring:
docker compose -f compose.yaml -f compose.override.yaml -f compose.host-network.yaml -f compose.host-processes.yaml up -d docker compose logs --tail 50
Rerun the same up -d after editing .env; a restart alone does not reload it. To update, pull, then up -d with the same files. If a pull is refused, make the silo-monitor package public on GitHub or docker login ghcr.io.
Set SILO_URL, SILO_API_KEY, and HOST_NETWORK_INTERFACE in your stack's environment variables. Add optional integrations as needed.
Complete Portainer stack
services:
silo-monitor:
image: ghcr.io/neilmc78/silo-monitor:latest
pull_policy: always
restart: unless-stopped
environment:
SILO_URL: ${SILO_URL:?Set SILO_URL}
SILO_API_KEY: ${SILO_API_KEY:?Set SILO_API_KEY}
SILO_MONITOR_API_KEY: ${SILO_MONITOR_API_KEY:-}
AUTH_MODE: ${AUTH_MODE:-silo}
PLEX_URL: ${PLEX_URL:-}
PLEX_TOKEN: ${PLEX_TOKEN:-}
IFD_URL: ${IFD_URL:-}
IFD_API_KEY: ${IFD_API_KEY:-}
SONARR_URL: ${SONARR_URL:-}
SONARR_API_KEY: ${SONARR_API_KEY:-}
RADARR_URL: ${RADARR_URL:-}
RADARR_API_KEY: ${RADARR_API_KEY:-}
ARR_REPLACEMENTS_ENABLED: ${ARR_REPLACEMENTS_ENABLED:-false}
REPLACEMENT_DATA_DIR: /data/replacements
PUSH_CONTACT: ${PUSH_CONTACT:-}
PUSH_DATA_DIR: /data/push
ACTIVITY_DATA_DIR: /data/activity
HOST_PROC_DIR: /host/proc
HOST_NETWORK_STATS_DIR: /host/network
LISTEN_ADDR: :8080
GOMAXPROCS: "1"
GOMEMLIMIT: 32MiB
volumes:
- notification-data:/data
- /proc:/host/proc:ro
- /sys/class/net/${HOST_NETWORK_INTERFACE:?Set HOST_NETWORK_INTERFACE}/statistics:/host/network:ro
ports:
- "${MONITOR_BIND:-127.0.0.1}:${MONITOR_PORT:-8091}:8080"
read_only: true
tmpfs:
- /tmp:size=1m,mode=1777
cap_drop:
- ALL
security_opt:
- no-new-privileges:true
pids_limit: 32
mem_limit: 48m
cpus: 0.10
volumes:
notification-data:Deploy the stack. Apply later configuration changes with Update the stack.
Keep /data writable and persistent to retain history, notifications, and replacement records. Back it up securely. Do not delete the volume or run docker compose down -v during updates.
When moving or renaming a stack, reuse its existing volume. For example:
volumes:
notification-data:
external: true
name: silo-performance_notification-dataReplace the example name with your actual volume name and keep the service mount as notification-data:/data.
Each version tag (v*) publishes static Linux binaries for AMD64 and ARM64 on the Releases page, with a systemd unit and an example configuration. Any Linux machine or container that can reach Silo works, including a small LXC container.
VERSION=v0.1.0 # a tag from the Releases page
ARCH=amd64 # or arm64
BASE=https://github.com/neilmc78/siloscope/releases/download/$VERSION
useradd --system --no-create-home --shell /usr/sbin/nologin siloscope
wget -O /usr/local/bin/siloscope "$BASE/siloscope-linux-$ARCH" && chmod 755 /usr/local/bin/siloscope
wget -O /etc/systemd/system/siloscope.service "$BASE/siloscope.service"
install -d -m 700 /etc/siloscope
wget -O /etc/siloscope/siloscope.env "$BASE/siloscope.env.example" && chmod 600 /etc/siloscope/siloscope.envEdit /etc/siloscope/siloscope.env, then systemctl daemon-reload && systemctl enable --now siloscope. Data is kept in /var/lib/siloscope. On a different machine from Silo, CPU, memory, disk and bandwidth come from Silo's own statistics and the top-processes panel is unavailable.
| Password | Two-step sign-in |
|---|---|
![]() |
![]() |
Siloscope asks you to sign in with a Silo administrator account. It sends the username and password to Silo's own sign-in, and for accounts with two-step sign-in, the authenticator or recovery code as well. Any account that isn't an administrator is refused. Siloscope then signs straight back out of Silo, keeps no password and no Silo session, and gives the browser its own session cookie for 30 days.
- Every 15 minutes Siloscope checks with Silo that each signed-in account still exists, is enabled and is still an administrator, and signs out any that isn't.
- Sign-in attempts are limited per name (5 wrong passwords in 15 minutes) and per address (20 in 10 minutes). Silo locks a name for 15 minutes after 10 failures, and that lock refuses even the right password everywhere, so Siloscope stops first: wrong passwords sent through Siloscope can't lock an account out of Silo. A correct password clears the count, as in Silo. Open from Silo still works while a name is locked here.
- Behind a reverse proxy (nginx, Tailscale serve or Funnel), Siloscope reads the visitor's address from
X-Forwarded-For, but only from proxies listed inTRUSTED_PROXIES. The default covers a proxy on the same machine:127.0.0.0/8,::1/128for the binary, plus Docker's private range172.16.0.0/12incompose.yaml, because a proxy on the host reaches the container from Docker's gateway address. SetTRUSTED_PROXIESyourself if your proxy runs on another machine. Siloscope passes the visitor's address on to Silo, so Silo's own per-address limits apply to the visitor rather than to Siloscope. - Requests that change anything are refused when they come from another website.
- Settings shows who is signed in, with a Sign out button.
- Plex sign-in and passkeys aren't supported here: sign in with the account's Silo password. An account that uses passkeys can add an authenticator app in Silo.
Two-step sign-in and break-glass accounts. Two-step sign-in applies to any administrator that signs in with a Silo password and isn't break-glass. Silo never asks a break-glass account for a second step, so it stays a password-only way back in, and the server Owner is break-glass by default. To protect Siloscope with two-step sign-in, keep the Owner as the emergency account and sign in to Siloscope with a second administrator that has a password and an authenticator app; only the Owner can make an account an administrator. An administrator that signs in to Silo with Plex has no Silo password and can't sign in to Siloscope.
If an authenticating reverse proxy already guards Siloscope, set AUTH_MODE=none to turn sign-in off.
| Admin sidebar | Plugin settings in Silo | Siloscope Settings |
|---|---|---|
![]() |
![]() |
![]() |
A small Silo plugin adds Siloscope under Plugin Apps in Silo's admin sidebar, and can sign you in to Siloscope with one click. It doesn't change Silo itself.
- From the latest release, download
siloscope-silo-plugin-linux-amd64(or-arm64) to match the machine Silo runs on. - In Silo, go to Admin → Integrations → Plugins, upload the file, and open Siloscope.
- Enter the address you use to open Siloscope in a browser, for example
https://silo.example.ts.net:8443. - For one-click sign-in, open Settings → Open from Silo in Siloscope, press Copy, paste the secret into Launch secret in Silo, and press Save.
Which address. Your browser opens it, not Silo, so 127.0.0.1 or localhost only works in a browser on the Silo machine itself. Use an address every device you use can reach: your Tailscale name with Siloscope's port works at home and away. When you open Silo over https, give Siloscope an https address too, or the browser warns before sending the sign-in.
How the one-click sign-in works. Silo tells the plugin who opened the link. Silo strips any such header a browser sends, and sets the administrator role only for an acting admin. The plugin signs a one-minute, single-use token with the shared secret, and your browser posts it to Siloscope. Siloscope checks the signature, the expiry and that the token hasn't been used, then asks Silo with its API key that the account is still an enabled administrator before opening a session. No password is involved. The secret is kept in Siloscope's database, or set it with SILO_LAUNCH_SECRET. Make a new secret stops the old one working at once. If the secrets don't match, Siloscope shows its sign-in page and says so.
Only Silo administrators see the link. It opens in the same tab, so Back returns to Silo. Silo marks the plugin Unverified because it comes from an upload rather than a plugin catalog. Check Connection says it isn't supported: Silo only runs connection checks for metadata and marker plugins (see docs/upstream/silo-notes.md). The plugin's source is in silo-plugin/.
Siloscope uses two keys:
| Key | Access | Used for |
|---|---|---|
SILO_API_KEY |
Full access | Everything that reads or changes Silo: health, sessions, history, users, libraries, transcoder settings, restarts and stopping streams |
SILO_MONITOR_API_KEY |
Selected scopes → admin:monitor:read only |
The live session events socket, which Silo only opens for this scope |
Create both under Admin → API Keys → Create Key in Silo. Builds without the scope choice on that page can create the monitoring key with the full key instead:
curl -sS -X POST "$SILO_URL/api/v2/admin/api-keys" \
-H "Authorization: Bearer $SILO_API_KEY" \
-H "Content-Type: application/json" \
-d '{"label":"Siloscope monitor","scopes":["admin:monitor:read"]}'Without the monitoring key, live Silo plays are picked up by a 15-second poll instead.
| Variable | Default | Purpose |
|---|---|---|
SILO_URL |
required | Silo base URL reachable by the monitor container |
SILO_API_KEY |
required | Silo admin API key |
SILO_MONITOR_API_KEY |
unset | Optional second Silo admin key with only the admin:monitor:read scope. Enables live session events for history; without it Siloscope polls Silo's session list every 15 seconds |
PLEX_URL |
unset | Optional Plex Media Server base URL, usually http://plex:32400 |
PLEX_TOKEN |
unset | Plex server-owner token; required with PLEX_URL (find your token) |
IFD_URL |
unset | Optional InfiniDysk backend URL, for example http://nzbdav:8080 on a shared private Docker network |
IFD_API_KEY |
unset | Same value as InfiniDysk's FRONTEND_BACKEND_API_KEY; required with IFD_URL, kept server-side |
SONARR_URL / SONARR_API_KEY |
unset | TV library integration; configure together, usually http://sonarr:8989 |
RADARR_URL / RADARR_API_KEY |
unset | Movie library integration; configure together, usually http://radarr:7878 |
ARR_REPLACEMENTS_ENABLED |
false |
Explicit opt-in to destructive Library actions; requires administrator-only proxy access |
REPLACEMENT_DATA_DIR |
/data/replacements in Docker/Compose |
Persistent replacement records; required when replacements are enabled |
AUTH_MODE |
silo |
silo requires signing in with a Silo administrator account; none turns sign-in off for use behind an authenticating proxy |
LISTEN_ADDR |
:8080 |
Address used by the Go server |
MONITOR_BIND |
127.0.0.1 |
Host address published by Compose |
MONITOR_PORT |
8091 |
Host port published by Compose |
HOST_NETWORK_INTERFACE |
eth0 |
Host NIC used by the optional bandwidth override |
HOST_NETWORK_STATS_DIR |
unset | Host network statistics path; supplied by the network override |
HOST_PROC_DIR |
unset | Host process statistics path; supplied by the process override |
PUSH_CONTACT |
unset | Enables Web Push when set to a real mailto: address or HTTPS contact URL |
PUSH_DATA_DIR |
/data/push in Compose |
Persistent notification data; required when push is enabled |
ACTIVITY_DATA_DIR |
/data/activity in Docker/Compose |
Playback history database (siloscope.db); set explicitly for standalone binaries |
The image is published for AMD64 and ARM64 at ghcr.io/neilmc78/silo-monitor:latest; release binaries are attached to each v* tag.
History is stored in SQLite at ACTIVITY_DATA_DIR/siloscope.db and kept indefinitely. Back up the data volume to keep it.
| Source | How plays are recorded |
|---|---|
| Silo | Silo's own finalized playback log is the authority for ended plays, with measured watch and pause time, client, device, IP and stream decisions. On first start Siloscope imports all of it. Live plays follow Silo's session events (with SILO_MONITOR_API_KEY) or a 15-second poll. |
| Plex | Siloscope listens on Plex's notification websocket. Each play records the source file (codec, resolution, HDR, bitrate) and the stream (decisions, hardware transcoding, subtitle burn-in, bandwidth, LAN/WAN). Watch and pause time are measured from Plex's position updates; seeks are not counted. A play counts as watched at 85%. On first start, Plex's own watch history is imported for earlier views; Plex keeps only finished views, so those rows have no duration or stream details. |
Silo posters are cached in the data directory the first time they are seen, so history keeps its artwork after Silo's signed image links expire.
Stats cover a week, month, year or all time, for Silo, Plex or both. A resumed play counts once, and unfinished attempts under 30 seconds are ignored, as in Tautulli. A show watched on both servers is one title. Concurrent streams are worked out from play start and end times, so imported history counts too. Weekday and hour use the viewing device's time zone.
Every Silo and Plex account belongs to a person. Siloscope links a Silo account and a Plex account to the same person automatically when either:
- the Silo account was signed in through the Plex sign-in plugin and carries the same plex.tv account ID, or
- both accounts have the same email address, ignoring case, unless they carry different plex.tv IDs.
Plex account names come from the Plex server; plex.tv supplies the IDs and emails, so allow outbound HTTPS to plex.tv. Silo doesn't verify the emails people enter, so check new matches on the Users page.
To fix a match by hand, open a person in Stats → Users. Link another account moves any other account to this person, Unlink splits an account off into its own person, and Rename changes the name shown. Automatic matching never undoes a manual change. The same actions are available through POST /api/people ({"action":"link","account_ids":[1,2]}, {"action":"unlink","account_id":1}, {"action":"rename","person_id":1,"display_name":"Sam"}), and GET /api/people lists people and their accounts.
Silo reports disk usage per mount. Library folders on one shared mount, such as a NAS share, are shown once and named after their Silo libraries. Each library's size and file count are totalled from Silo's catalog every six hours, without scanning the share.
The Nodes tab shows Silo's remote stream nodes: transcode workers that convert streams live on another machine, and proxies that relay streams to viewers. A single-server Silo has none; hide the tab under Settings → Navigation.
- Appearance: theme, accent colour (presets or any colour) and animated effects: Console scanlines and cursor, Matrix code rain. Reduced-motion devices always get still versions.
- Navigation: reorder the bottom tabs or hide the ones you don't use. Settings always stays.
- System, Playing, Nodes: choose which panels and details appear, reorder System panels, and set chart height and time range.
- Transcoder: Silo's transcoding and 4K permissions, hardware acceleration, HDR tone mapping, buffering, throttling and playback routing.
- Restart: see pending restart requirements and restart Silo after confirmation.
- Notifications: push alerts; see below.
Display settings are saved per device.
Choose the alerts you want under Settings > Notifications:
| Alert | Scenario |
|---|---|
| CPU | Sustained high usage and recovery, with configurable thresholds and cooldown |
| Disk | Usage exceeds your selected threshold |
| Playback | A new Silo or Plex session starts |
| Transcoding | A new session is transcoding at its one-time check around one minute after discovery; includes audio-only transcoding |
| Service | Silo, Plex, or an enabled node becomes unavailable or recovers |
Set PUSH_CONTACT, open Siloscope over HTTPS, and enable notifications on each device. On iOS 16.4 or later, first add Siloscope to the Home Screen and open it there. Titles and usernames are optional notification details.
Alerts can arrive with the dashboard closed. Delivery depends on connectivity and device permissions; Siloscope cannot alert while it is down.
InfiniDysk uses the private backend (normally port 8080), not its port 3000 browser interface. Do not publish its backend port for this integration. Generate a shared secret with openssl rand -hex 32, set it as FRONTEND_BACKEND_API_KEY in the InfiniDysk stack and as IFD_API_KEY in Siloscope, then recreate both services. Recreating IFD may briefly interrupt streams or downloads. Keep the existing normal IFD API key unchanged: that key supports statistics but cannot authenticate the active-file WebSocket in IFD 1.5.0.
For Dockhand or Portainer, supply the generated value through each stack's environment configuration. Both containers must share a private Docker network. In the Siloscope service, use:
environment:
IFD_URL: http://nzbdav:8080
IFD_API_KEY: ${IFD_API_KEY}Replace nzbdav with your IFD service's network name if different. In the IFD service, map FRONTEND_BACKEND_API_KEY: ${FRONTEND_BACKEND_API_KEY} under environment, supplying the same generated value. Keep secrets out of Git. Leave both Siloscope IFD variables unset to disable the integration.
Throughput is sampled every five seconds; fetched throughput uses IFD's recorded byte totals and may lag while those totals are persisted. Served throughput uses its cumulative byte counter. Per-file fetching speed is derived from live read updates; unknown values appear as --, not zero.
Under Settings > System, move the InfiniDysk panel up or down, toggle its individual elements, or change the shared graph height and time window. These preferences are saved per device. Existing saved panel orders are preserved, with InfiniDysk appended until repositioned.
Library searches titles already managed by Sonarr or Radarr; it does not add new titles. Search is available without enabling replacements.
| Search | Result |
|---|---|
| Movie or series title | Matching library items |
the simpsons s05e19 |
The exact episode |
the simpsons s05 |
Episodes from season five |
If multiple series match, choose one to continue.
Enable ARR_REPLACEMENTS_ENABLED=true to use Replace. Review the file and release, then confirm Delete and search to delete the file, blocklist its release, and search for a replacement. Recent operations show progress and offer Retry search when available.
An individually stored episode from a season pack can be replaced: only that episode's file is deleted, the whole pack release is blocklisted, and a replacement search is requested for that episode.
Replacement can interrupt playback and does not guarantee a new or better file. Files containing multiple episodes and files without identifiable download history must be handled directly in Arr. If an operation is uncertain, check Arr before retrying; do not clear replacement records to bypass the warning.
Select one or more Silo or Plex sessions in Playing, then choose Terminate. This acts immediately without a confirmation dialog.

















