-
Notifications
You must be signed in to change notification settings - Fork 0
Intel
The Intel tab is a read-only lifetime performance dashboard. It answers the question "how is Nudgarr performing for my library overall?" — distinct from the Sweep tab which covers the last run. The longer Nudgarr has been running, the richer the data becomes.
Intel is a desktop-only feature. It is not available in the mobile UI.
Open the Intel tab in the desktop navigation bar, between Imports and Notifications.
On a fresh install, or after a Reset Intel, Intel shows a Building… state in the Library Score until enough data has accumulated. The threshold is 10 confirmed imports or 30 completed sweep runs across all instances, whichever comes first. All other cards still display data where available — only the score itself waits for the minimum threshold.
A single 0–100 number summarising overall Nudgarr health. It is a weighted composite of four sub-scores:
| Component | Weight | What it measures |
|---|---|---|
| Success Rate | 40% | Percentage of tracked items that have been confirmed imported |
| Turnaround | 25% | How quickly items are importing — shorter is better |
| Stuck Items | 20% | How many items are repeatedly searched with no result |
| Sweep Efficiency | 15% | How much of the eligible library is being covered per sweep |
The score is rated as follows:
| Score | Label |
|---|---|
| 70–100 | Good Health |
| 40–69 | Needs Attention |
| 0–39 | Poor Health |
The four mini-bars beneath the ring show each sub-score individually so you can see which component is driving the overall number.
Lifetime performance metrics across all instances combined.
| Metric | What it shows |
|---|---|
| Success Rate | Percentage of all tracked items that have been confirmed imported at least once. Includes items still pending import in the denominator. |
| Average Turnaround | Average number of days from first search to confirmed import across all confirmed imports. |
| Searches per Import | Average search count at the moment of confirmed import. May be understated for titles that were previously auto-excluded and had their count reset before importing. |
| Stuck Items | Count of items searched at or above your auto-exclusion threshold with no confirmed import. Shows N/A if auto-exclusion thresholds are both set to 0. |
| Cutoff Unmet / Backlog split | How confirmed imports break down by search type — only counts first-time imports, not quality upgrades on the same title. |
| Quality Upgrades Confirmed | Total confirmed quality upgrade events — items that were already imported and later found at a higher quality. Excludes first-time acquisitions. |
A table showing lifetime metrics per instance. Columns are:
| Column | What it shows |
|---|---|
| Sweep Runs | Total sweep runs recorded for this instance — already protected by sweep_lifetime and unaffected by any clear operation. |
| Total Searched | Cumulative items searched across all runs. Counts each search event including repeat searches of the same item. |
| Confirmed Imports | Unique titles confirmed imported at least once for this instance. Does not count quality upgrades on the same title. |
| Success Rate | Confirmed imports divided by unique items tracked in search history for this instance. |
| Avg Turnaround | Average days from first search to confirmed import for this instance. |
| Eligible Used | Lifetime average ratio of items searched vs items eligible per sweep. Shown as a bar — amber when consistently at or near 100%, which may indicate the search cap is too low. |
| Stuck Items | Items at or above the auto-exclusion threshold with no confirmed import for this instance. |
All per-instance metrics other than Sweep Runs and Total Searched come from the protected aggregate and survive Clear History and Clear Stats.
A list of individual titles that Nudgarr has been repeatedly searching with no confirmed import and no auto-exclusion applied yet. Sorted by search count descending — the most-searched items appear first.
Each row shows the title, instance, search count, first searched date, and library added date (shown as Unknown if not recorded).
This is the only actionable card in Intel. Use it to decide which titles to manually exclude, or as a signal to review your auto-exclusion threshold. Stuck items are a live query — they update immediately as items are excluded or imported.
If auto-exclusion thresholds are both set to 0, this card shows a message that thresholds must be enabled to surface stuck items.
Lifetime summary of your exclusions activity.
| Metric | What it shows |
|---|---|
| Total Exclusions | Current count of all exclusion rows, manual and auto combined. |
| Auto this month | Count of auto-exclusions added in the current calendar month. |
| Manual / Auto split | Breakdown of exclusions by how they were added. |
| Avg searches at auto-exclusion | Average search count captured at the moment each title was auto-excluded. This is a point-in-time snapshot taken at exclusion time — unaffected by any subsequent count resets. |
The calibration signal answers: "of all the titles Nudgarr auto-excluded and then later gave another chance to, how many actually ended up importing?"
It shows a count ("11 of 45") and a plain-English assessment:
| Ratio | Assessment |
|---|---|
| Above 20% | Your threshold may be excluding titles too early. Consider raising it. |
| 5–20% | Your threshold appears well-calibrated. |
| Below 5% or insufficient data | Not enough auto-unexclude cycles to draw conclusions yet. |
The signal requires completed exclude-then-unexclude cycles to have any data. On a fresh install, or after Reset Intel, it shows a cold start message until cycles have accumulated. Both the auto-unexclude timer and manual deletion of auto-excluded titles count as giving a title a second chance.
A bar chart showing import success rate broken down by how long items had been in the library at the time Nudgarr first searched them.
| Bucket | Age at first search |
|---|---|
| Under 1 month | Less than 30 days |
| 1 to 3 months | 30 to 89 days |
| 3 to 6 months | 90 to 179 days |
| 6 to 12 months | 180 to 364 days |
| 12+ months | 365 days or more |
Bars are colour-coded: green above 70%, blue 50–70%, amber below 50%.
Items where the library added date was not recorded cannot be bucketed. The count of unbucketed items is shown below the chart as a note rather than a bar, so the chart scale is not distorted.
What to look for: A steep drop in success rate at the 3–6 month or 6–12 month buckets typically means long-tail content that indexers do not carry. A consistent rate across all buckets means your indexer coverage is strong.
Shows how many titles were imported once versus upgraded at least once after the initial import.
| Metric | What it shows |
|---|---|
| Imported Once | Titles confirmed imported exactly once — no upgrades on record. |
| Upgraded | Titles confirmed imported more than once — at least one quality upgrade occurred. |
| Most Common Upgrade Paths | The single most frequent quality_from → quality_to upgrade pair, shown separately for Radarr and Sonarr. Only genuine upgrades (where a prior quality is recorded) are included — first-time acquisitions are excluded. |
Radarr and Sonarr quality profile naming conventions differ, which is why paths are shown per app rather than combined.
Per-instance bars showing the lifetime average ratio of items searched vs items eligible. This is the same data as the Eligible Used column in the Instance Performance table, given more space with supporting text.
An amber callout appears beneath an instance bar when it has been consistently hitting its search cap across many runs. The callout reads: "This instance is consistently hitting its search cap. If your library is growing, lower-priority items may wait longer between searches. Consider raising your search limit."
Lifetime average note: All efficiency figures are lifetime averages across all runs for that instance. If you recently raised your search cap, the bar will drift toward the new reality over time but will not immediately reflect it. The bar label states "lifetime average" to make this clear.
Intel is designed to survive the operations that erase other data.
| Operation | Effect on Intel |
|---|---|
| Clear History | No effect on Intel data |
| Clear Stats | No effect on Intel data |
| State Retention pruning | No effect on Intel data |
| Reset Intel (Danger Zone) | Clears all Intel data — the only operation that does so |
Intel data accumulates in two protected database tables: intel_aggregate (one row that grows with every confirmed import) and exclusion_events (one row per exclude or unexclude action). Neither is touched by any normal clear or prune operation.
The search count for each confirmed import is snapshotted at the moment the import is confirmed — before any future auto-unexclude reset can alter it — so the searches-per-import metric remains accurate even on installs with heavy auto-exclusion cycling.
Advanced → Danger Zone → Reset Intel
Permanently clears all Intel data: resets intel_aggregate to zero defaults and deletes all rows from exclusion_events. Intel data begins accumulating again immediately on the next sweep. A confirmation dialog appears before the action runs.
This does not affect search history, confirmed imports, exclusions, sweep lifetime counters, or any other data.