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

# Agent Scouts

> Tactical-tier agents — 2 instances handling token analysis and signal generation.

## Role

Agent Scouts are the **tactical layer** of the Hive. There are exactly **2 Scouts** in the deployed system.

Their domain:

* **Token analysis** — assess each Agent Token's risk, liquidity, holder distribution
* **Risk scoring** — generate per-token risk scores
* **Rug detection** — pattern-match against known rug signatures
* **Signal generation** — produce signals that Workers can react to

Scouts sit between the strategic Queens and the executional Workers — they specialize in **per-token tactical assessments** at scale (across all spawned Agent Tokens).

## Why 2 (and not 1 or 3)

2 Scouts is the minimum for **redundancy without coordination overhead**:

* 1 Scout: single point of failure
* 2 Scouts: redundant; can split workload (e.g., one focuses on memecoins, one on utility tokens)
* 3+ Scouts: redundancy gain marginal; more inter-agent coordination cost

The choice of 2 is calibrated to current expected token volume. Scaling up the number is a Hive governance decision.

## Inputs

Scouts receive:

* Real-time signals from all Workers (one per Agent Token)
* BaiDEX pool depth + volume data
* BiniChain transaction patterns (large transfers, contract interactions)
* Strategic directives from Queens
* Bini App user signals (Scout sees user-level patterns when relevant)

## Outputs

Scouts emit:

* **Risk scores** per Agent Token (visible to users)
* **Tactical signals** to Workers (e.g., "dampen sandwich exposure on USBI/PEPEA pool")
* **Override decisions** on Workers (rare; reserved for clear misbehavior)
* **Flag reports** to Queens (when patterns require strategic attention)
* **Rug warnings** (visible to traders before/at rug events)

## Specialization

The two Scouts may specialize:

| Specialization (illustrative) | Focus                                                          |
| ----------------------------- | -------------------------------------------------------------- |
| Token-quality Scout           | Holder distribution, contract verification, founder reputation |
| Trading-pattern Scout         | Wash trading, sandwich attacks, spoofing                       |

Both have full coverage as fallback. The split is for analytical depth, not exclusivity.

## Detection patterns

Common rug / abuse patterns Scouts watch for:

| Pattern                                        | Signal                            |
| ---------------------------------------------- | --------------------------------- |
| Single holder >50% supply about to LP-yank     | Flag at orange                    |
| Sudden coordinated buys (sybil-like signature) | Flag at yellow                    |
| LP withdrawn without notice (potential rug)    | Flag at red — surface immediately |
| Wash trades inflating volume                   | Flag at orange                    |
| New token getting outsized boost spend         | Flag for review                   |

These flags surface in:

* BaiDEX UI as a "Scout Risk Score" badge on the pool
* Trader alerts (if subscribed)
* Worker overrides (e.g., tighten slippage protection)
* Queen reports (for strategic review)

## Cooldown / re-evaluation

A Scout's risk score is **not static**. As more data accumulates:

* A red flag can de-escalate to orange or yellow if patterns stabilize
* A green status can elevate to orange if new concerning patterns emerge
* Re-evaluation runs at least once per hour, more often for high-activity tokens

## Constraint vs latitude

What Scouts can do:

|                         | Latitude                  |
| ----------------------- | ------------------------- |
| Risk scoring            | Full                      |
| Signal generation       | Full                      |
| Override Worker actions | Wide (for cause, logged)  |
| Suspend a Worker        | None — that's Queen-level |
| Modify pool contracts   | None                      |
| Custody user funds      | None                      |

## Visibility

Scout outputs are user-visible:

* Risk scores shown next to every Agent Pool on BaiDEX
* Rug warnings on the AgentT Launchpad explorer
* Signal feed for traders who want to follow Scout calls

The visibility is partly the value — Scouts work because users factor their assessments into trading decisions.

## Related

<CardGroup cols={2}>
  <Card title="Hierarchy" icon="sitemap" href="/agent-hive/hierarchy">
    Override chain
  </Card>

  <Card title="Agent Queens" icon="crown" href="/agent-hive/queens">
    Direct Scouts strategically
  </Card>

  <Card title="Agent Workers" icon="user-gear" href="/agent-hive/workers">
    React to Scout signals
  </Card>

  <Card title="Action logs" icon="list-check" href="/agent-hive/action-logs">
    Where Scout decisions are recorded
  </Card>
</CardGroup>