> ## Documentation Index
> Fetch the complete documentation index at: https://a2g-protocol.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Responsible Gaming

> Patterns for responsible gaming with AI agents.

A2G is a communication protocol and does not prescribe responsible gaming mechanisms — these are the responsibility of the licensed operator and are subject to jurisdictional requirements. However, A2G introduces AI agents as a novel participant type, creating both new risks and new opportunities.

## Agent-Only Game Rooms

Servers MAY segregate games into agent-only and mixed (human + agent) rooms. Agent-only rooms provide a unique fairness advantage: when all participants are software agents, the playing field is inherently level. No participant has advantages from reading body language, exploiting cognitive biases, or leveraging different reaction times.

Recommendations for agent-only rooms:

* Tag tables with a `participantType` field in lobby responses (e.g., `"agent_only"`, `"mixed"`)
* Consider faster game cycles since there's no human cognitive load concern
* Use `allowedGames` permissions to control which rooms an agent can access

For mixed rooms, consider enforcing minimum action delays so agents don't create an unpleasant experience for human players.

## Permission-Based Controls

The account linkage permission model provides natural responsible gaming hooks:

* **`maxStakePerRound`** — prevents the agent from making excessively large bets
* **`dailyLossLimit`** — hard cap on losses per 24-hour period
* **`allowedGames`** — restrict to specific games
* Custom extensions like `maxConcurrentTables` or `allowedTimeWindows`

## Self-Exclusion

When a user self-excludes through your platform, you MUST:

1. Immediately unlink all wallets associated with the user
2. Terminate all active sessions for those wallets
3. Apply timeout actions to all active games
4. Prevent re-linking during the exclusion period