Skip to content

Configuration

Jump to bottom
o51r15 edited this page Aug 22, 2026 · 9 revisions

All configuration lives in config.yaml. The repo ships a fully annotated config.example.yaml — copy it and edit. Most options are also editable from the Web UI Settings page, which writes back to the same file. The scheduler reloads config.yaml from disk before every scan, so changes take effect on the next cycle without a restart. The one exception is web.port, which requires a restart.

Connections

Torrent Client

Inspectarr supports three torrent clients. Set torrent_client to choose which one to use, then configure the matching block. Only one client is active at a time. The Settings UI shows fields for the active client only.

torrent_client: qbittorrent   # qbittorrent | transmission | deluge

qbittorrent:
  url: http://192.168.1.100:8080
  username: admin
  password: changeme

transmission:
  url: http://192.168.1.100:9091
  username: ""
  password: ""

deluge:
  url: http://192.168.1.100:8112
  password: ""
Setting Purpose
torrent_client Which torrent client to use: qbittorrent (default), transmission, or deluge
qbittorrent.url qBittorrent Web UI URL including port
qbittorrent.username / password qBittorrent Web UI credentials
transmission.url Transmission RPC URL (default port 9091)
transmission.username / password Transmission RPC credentials (optional — leave empty if no auth)
deluge.url Deluge Web UI URL (default port 8112)
deluge.password Deluge Web UI password (default: deluge)

*arr Apps

arrs:
  sonarr:
    enabled: true
    url: http://192.168.1.100:8989/sonarr     # include the base path if you use one
    api_key: your_api_key
  radarr:
    enabled: true
    url: http://192.168.1.100:7878/radarr
    api_key: your_api_key
  lidarr:
    enabled: false
    url: ~
    api_key: ~
Setting Purpose
arrs.<app>.enabled Whether this *arr client is active
arrs.<app>.url The *arr URL, including any base path (e.g. /radarr)
arrs.<app>.api_key Found in the *arr under Settings → General

Rules

Rules decide what counts as a bad torrent. Each rule watches one qBittorrent category and ties it to one *arr app. See Rules for the full breakdown.

rules:
  - name: "TV Bad Extensions"
    category: "tv-sonarr"
    app: sonarr
    conditions:
      match_mode: any
      bad_extensions: [".exe", ".bat", ".msi", ".scr", ".lnk"]
      min_file_size_mb: ~
      bad_filename_patterns: []

Scanning

Controls how and when Inspectarr checks for new torrents. Polling and webhooks are independent — you can use either or both.

scanning:
  polling:
    enabled: true
    interval_seconds: 300
  webhooks:
    enabled: false
    secret: ""
    scan_delay_seconds: 60
Setting Purpose
scanning.polling.enabled Whether the scheduler polls on an interval (default true)
scanning.polling.interval_seconds Seconds between scheduled scans (default 300)
scanning.webhooks.enabled Accept incoming webhook push events from *arr apps (default false)
scanning.webhooks.secret Shared secret for HMAC validation — leave empty to skip validation
scanning.webhooks.scan_delay_seconds Delay after receiving a webhook before scanning, to let the torrent connect to the swarm (default 60)

When webhooks are enabled, Inspectarr listens on POST /webhook/sonarr, /webhook/radarr, and /webhook/lidarr. Configure your *arr app's Connect settings to point at these URLs. The webhook URLs are shown on the Scheduler page.

Backward compatibility: If you have poll_interval_seconds at the top level from an older config, it still works — the value is used as the default for scanning.polling.interval_seconds.

Behavior

on_arr_failure: delete
dry_run: false
Setting Purpose
on_arr_failure delete removes the torrent from qBittorrent even if the *arr blocklist call fails; abort skips deletion and queues a retry
dry_run true logs matches but takes no action — ideal for testing a new rule

Retry

retry:
  enabled: true
  max_attempts: 10
  interval_seconds: 600

When an *arr call fails and on_arr_failure is abort, the torrent is queued and retried up to max_attempts times, interval_seconds apart.

Notifications

Inspectarr uses Apprise for notifications, supporting Pushover, Telegram, Discord, email, and 100+ services. See Notifications for setup details and URL examples.

notifications:
  apprise:
    enabled: true
    urls:
      - pover://user@token              # Pushover
      - tgram://bottoken/ChatID         # Telegram
      - discord://webhook_id/token      # Discord
    notify_on: [action, error, startup, dry_run]
  digest:
    enabled: false
    use_ollama: false
  summary:
    enabled: false
    schedule: daily
    use_ollama: true
Setting Purpose
notifications.apprise.enabled Master switch for notifications
notifications.apprise.urls List of Apprise URLs — one per notification service
notifications.apprise.notify_on Event types that trigger a notification: action, error, startup, dry_run
notifications.digest.enabled Batch notifications into a single digest instead of sending individually
notifications.digest.use_ollama Use Ollama to generate a natural-language digest summary
notifications.summary.enabled Generate periodic summaries of recent activity
notifications.summary.schedule daily or weekly
notifications.summary.use_ollama Use Ollama for natural-language summary narration

Prowlarr

Enables indexer health scoring, auto-reorder, auto-manage, and grab attribution. See Prowlarr Indexer Scoring for the full explanation of each weight.

prowlarr:
  enabled: true
  url: http://192.168.1.100:9696/prowlarr
  api_key: your_api_key
  base_priority: 10
  reorder_interval_hours: 24
  min_grabs_before_scoring: 10
  scoring:
    response_time_weight: 0.25
    failure_rate_weight: 0.30
    malicious_weight: 0.20
    grab_success_weight: 0.25
    auth_failure_mult: 3.0
    grab_failure_mult: 2.0
    query_failure_mult: 1.0
    rss_failure_mult: 0.5
    backoff_penalty: 20
    malicious_penalty_per_hit: 10
Setting Purpose
prowlarr.enabled Master switch for indexer health scoring
prowlarr.url Prowlarr URL including any base path
prowlarr.api_key Prowlarr API key (Settings → General)
prowlarr.base_priority Priority assigned to the top-ranked indexer; others count up from here
prowlarr.reorder_interval_hours Hours between automatic priority reorders (default 24)
prowlarr.min_grabs_before_scoring Minimum history records before an indexer is scored (default 10)

Scoring weights

Four weights that must sum to 1.0:

Weight Default Measures
response_time_weight 0.25 Average API response time
failure_rate_weight 0.30 Weighted failure rate across failure types
malicious_weight 0.20 Rate of malicious content served
grab_success_weight 0.25 Ratio of successful grabs to total grabs

Failure type multipliers control how heavily each failure type counts toward the failure rate sub-score:

Multiplier Default Severity
auth_failure_mult 3.0 Auth failures — most severe
grab_failure_mult 2.0 Grab failures
query_failure_mult 1.0 Query failures (baseline)
rss_failure_mult 0.5 RSS failures — least severe

Additional penalties:

Setting Default Purpose
backoff_penalty 20 Points deducted if the indexer is currently in Prowlarr backoff
malicious_penalty_per_hit 10 Points deducted per confirmed malicious torrent

Ollama (AI scoring)

When configured, Ollama analyzes indexer data and produces an AI-powered health score. If Ollama is unreachable or not configured, the deterministic formula above is used automatically.

  ollama:
    url: http://192.168.1.125:11434
    model: gemma3:latest
    timeout: 120
    cache_ttl_hours: 24
Setting Default Purpose
ollama.url "" Ollama API endpoint. Blank disables every AI feature.
ollama.model "" Model name — must pass validation before it can be selected
ollama.timeout 120 Seconds — hard cutoff per scoring run. 300 is safer if models load cold.
ollama.cache_ttl_hours 24 How long an AI result is reused. The cache key includes the model and system prompt, so changing either always rescores.
ollama.update_check_hours 24 How often to check whether the model behind this name has been replaced. 0 disables.
ollama.system_prompt "" Custom scoring prompt. Empty uses the built-in default.

All of these are editable under Settings → AI, which also validates a model before letting you use it. See AI & Model Validation.

Context size matters. The scoring prompt is roughly 5,700 tokens at 39 indexers. A 4k-context model cannot hold it and will return fabricated scores rather than an error. 8k is the practical minimum; 16k is comfortable.

Auto-manage

Automatically disables indexers that consistently score below a threshold and re-enables them after a cooldown. Auto-manage runs after every scan cycle independently of the reorder interval.

  auto_manage:
    enabled: false
    disable_threshold: 30.0
    consecutive_runs: 3
    cooldown_hours: 24
Setting Default Purpose
auto_manage.enabled false Enable automatic disable/re-enable
auto_manage.disable_threshold 30.0 Health score below which an indexer is flagged
auto_manage.consecutive_runs 3 Must score below threshold this many consecutive cycles before being disabled
auto_manage.cooldown_hours 24 Hours before a disabled indexer is automatically re-enabled

You can always manually override from the Indexers page.

Remediation

Controls what happens when a rule matches. Every finding is graded, the worst grade wins, and two thresholds split that grade into three outcomes. Full explanation in Quarantine & Severity.

remediation:
  min_severity: LOW                    # act at or above this
  remediate_at: LOW                    # delete at or above this
  quarantine_timeout_minutes: 0        # 0 = hold indefinitely
  quarantine_timeout_action: release   # "release" or "remediate"
  severity_overrides: {}
Setting Default Purpose
remediation.min_severity LOW Lowest severity that is acted on. Anything below is recorded only.
remediation.remediate_at LOW Lowest severity that is deleted. Between the two, catches are quarantined.
remediation.quarantine_timeout_minutes 0 How long a hold lasts. 0 holds indefinitely.
remediation.quarantine_timeout_action release What an elapsed timeout does — release resumes, remediate deletes.
remediation.severity_overrides {} Reclassify an extension or signal. Extension keys win over signal keys.

Both thresholds default to LOW, which means everything flagged is deleted immediately and nothing is ever quarantined — the behaviour Inspectarr has always had. Raise remediate_at to open a review band.

  severity_overrides:
    ".rar": LOW                   # RAR is routine in this library
    ".iso": CRITICAL              # a disc image never is
    "bad_filename_pattern": LOW   # trust these patterns less

The thresholds and timeout are editable under Settings → Rules → Remediation & Severity. severity_overrides has no form field — edit it in the Raw YAML view.

Web UI

web:
  port: 8585
  scheduler_autostart: false
  auth:
    enabled: false
    username: admin
    password: changeme
Setting Default Purpose
web.port 8585 HTTP port for the web UI (requires restart to change)
web.scheduler_autostart false Start the scan scheduler automatically when the web UI launches
web.auth.enabled false Password-protect the web UI
web.auth.username admin Login username
web.auth.password changeme Login password

Advanced

These settings rarely need changing.

logging:
  log_file: ./data/inspectarr.log.json
  retention_days: 30
  level: INFO

state:
  db_file: ./data/inspectarr.db
Setting Default Purpose
logging.log_file ./data/inspectarr.log.json Path to the JSON Lines log file
logging.retention_days 30 Days before old log entries and state records are pruned
logging.level INFO DEBUG for verbose output
state.db_file ./data/inspectarr.db Path to the SQLite state database

GitHub · Releases · Issues

Getting started

How it works

Using it

Help

Clone this wiki locally