> ## Documentation Index
> Fetch the complete documentation index at: https://hashbot.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Vision Insight

> Cross-server threat intelligence that shares ban data across the Vision network — automatically flagging or banning known scammers when they join your server.

**Vision Insight** is a cross-server threat intelligence system built into Hashbot Vision. When a scammer is banned on any participating server, their ID enters a shared threat pool. When that user joins your server, Vision Insight alerts your moderators or automatically bans them — before they can cause damage.

## How it works

<Steps>
  <Step title="A ban happens anywhere in the network">
    When a user is banned on any server running Vision — whether by a moderator, a moderation bot, or Vision's own detection — the ban is reported to the Vision Insight pool with contextual metadata.
  </Step>

  <Step title="The pool assigns a confidence score">
    Each reported user accumulates a threat score based on the number of independent reports, the source reliability, the ban reasons, and other signals. Higher scores indicate stronger evidence of malicious behavior.
  </Step>

  <Step title="The user joins your server">
    When a flagged user joins a server with Vision Insight enabled, the system checks them against the pool and takes action based on your configured protection mode.
  </Step>
</Steps>

## Protection modes

| Mode           | Behavior                                                                                                    |
| -------------- | ----------------------------------------------------------------------------------------------------------- |
| **Protect**    | Automatically bans users above your auto-ban threshold and sends an alert for lower-confidence matches      |
| **Alert Only** | Sends an alert to your log channel — no automatic action is taken                                           |
| **Feed Only**  | Your server contributes ban data to the pool but does not receive alerts or take action on incoming members |

<Info>
  **Protect** mode is recommended for most servers. It auto-bans only high-confidence threats and alerts you for everything else, giving you full control over borderline cases.
</Info>

## Confidence scoring

Every flagged user in the pool has a **confidence score** from 0 to 100, reflecting how likely they are to be a genuine threat. The score is calculated from multiple signals:

* **Number of independent reports** — more servers banning the same user increases confidence
* **Source reliability** — bans from verified high-signal servers carry more weight
* **Detection context** — bans triggered by automated scam detection score higher than unexplained bans
* **Bot intelligence** — bans performed by recognized moderation bots contribute signal

<Warning>
  Exact scoring weights and thresholds are intentionally not published to prevent gaming. You can tune the **minimum confidence** and **auto-ban threshold** to match your server's risk tolerance.
</Warning>

## Cross-bot intelligence

Vision Insight doesn't only capture bans from Hashbot. When any moderation bot bans a user on a participating server, Vision Insight detects the ban event and adds it to the pool. This means the network benefits from the enforcement actions of your entire moderation stack.

## Anti-poisoning safeguards

The Vision Insight pool includes multiple layers of protection against poisoning attacks — where a bad actor might try to mass-report legitimate users to trigger false bans. These safeguards ensure the integrity of the threat pool even at scale.

<Info>
  If you believe a user has been incorrectly flagged, use `/vision-insight-check` to view their threat report details.
</Info>

## Commands

| Command                 | Description                                                                                |
| ----------------------- | ------------------------------------------------------------------------------------------ |
| `/vision-insight`       | Configure Vision Insight (enable/disable, set mode, log channel, thresholds)               |
| `/vision-insight-check` | Check if a specific user is flagged in the threat pool                                     |
| `/vision-insight-stats` | View pool statistics (total threat IDs, contributing servers, your server's contributions) |

## Pool statistics

The Vision Insight pool is actively maintained and growing:

* **221,000+** threat IDs tracked
* **27** contributing servers
* Continuous ingestion from bans across the network

<Tip>
  Use `/vision-insight-stats` to see live pool statistics and your server's contribution to the network.
</Tip>

## Configuration via Dashboard

All Vision Insight settings are available on the [Vision Dashboard](https://dashboard.hashbot.com) under the **Vision Insight** tab, including protection mode, confidence thresholds, and log channel selection.