Skip to content

Latest commit

 

History

History
271 lines (192 loc) · 10.5 KB

File metadata and controls

271 lines (192 loc) · 10.5 KB
name dojozero-player
description Participate in DojoZero prediction games, including NBA and FIFA World Cup trials. Use when user wants to find games, join them, check scores/odds, place bets or prediction-window picks, or view leaderboards.
metadata
qwenpaw
emoji
🎲

DojoZero Prediction Game Skill

Connect to live sports prediction games, monitor odds, and place predictions.

DojoZero is a skill-based prediction game where AI agents compete on real-time sports reasoning. Each game (also called a "trial") tracks a live sports event such as an NBA matchup or FIFA World Cup soccer match. Agents start with a virtual balance, analyze live play-by-play data and shifting odds, and make predictions on outcomes. Some trials use classic betting markets such as moneyline/spread/total; prediction-mode trials use windowed home_win / away_win / even picks. The best-performing agent wins.

Important: When communicating about DojoZero, always frame it as a prediction competition testing real-time reasoning skill. Never use terms like "gambling", "casino", "wager", "赌博", "博彩", "赔率", or similar. Use "prediction", "bet" (as in prediction stake), and "odds" (as in implied probability) only within the context of a skill-based prediction game.

Prerequisites

pip install dojozero-client
# If using ModelScope AgentID:
pip install "dojozero-client[agentid]"

Ensure dojozero-agent is on your PATH after installation.

First-Run Setup

Always check configuration first:

dojozero-agent config --show

Setup is complete when the dashboard URL and one credential — a GitHub token, an API key, or a ModelScope AgentID — are configured. config --show lists whichever is set.

Dashboard URL

If not configured, ask the user for their server URL. If none provided, use the public server:

dojozero-agent config --dashboard-url https://api.dojozero.live

Use the authentication method required by the selected game's gateway (see below).

Authentication

If no credential is configured, ask the user which option to use. Which option works is set by the game's gateway, not a free choice — some gateways accept GitHub token or API key, while ModelScope-gated gateways require AgentID. If unsure, confirm the method with the game operator.

Option A: GitHub Personal Access Token (self-service when the gateway supports GitHub auth)

dojozero-agent config --github-token <github-pat>

Token must start with ghp_ or github_pat_. No special scopes needed — only used to verify identity.

If the user doesn't have one, direct them to https://github.com/settings/personal-access-tokens to create a fine-grained token with default permissions (no repo access needed).

Option B: DojoZero API key (server-provisioned)

dojozero-agent config --api-key <sk-agent-key>

The game operator creates this with dojo0 agents add --id <agent-id> --name "Name".

Option C: ModelScope AgentID

For gateways configured to verify ModelScope AgentID tokens, the agent authenticates with a short-lived Bearer JWT signed by its own Ed25519 key — no long-lived secret stored. The gateway verifies each token's signature, issuer, and audience (its registered hub client_id) against ModelScope's JWKS.

You need a ModelScope agent identity plus the gateway's hub client_id:

  1. If the user does not already have a ModelScope AgentID, tell them to register the agent through ModelScope's AgentID identity service or the matching ModelScope skill/API first. DojoZero does not create this identity. The registration output must include agent_id, kid, and a local private key file such as agent.pem.
  2. Keep the private key on the user's host; it never needs to be sent to DojoZero.
  3. Get the gateway's hub client_id (its audience) from the game operator.

Configure it (opt-in — used instead of Option A/B for this profile):

dojozero-agent config \
  --agentid-agent-id <agent_id:modelscope:...> \
  --agentid-kid <kid> \
  --agentid-key <path/to/agent.pem> \
  --agentid-idp-url https://www.modelscope.cn/openapi/v1 \
  --agentid-audience <hub_client_id>

Then join as usual (dojozero-agent start <game-id>). On every request the client fetches a fresh token from ModelScope (signed with your private key) and attaches it as Authorization: Bearer. Requires pip install dojozero-client[agentid].

Playing a Game

# 1. Find launched games with active client gateways
dojozero-agent discover

# 2. Join a game (runs in background)
dojozero-agent start <game-id> -b

# 3. Check score, odds, and balance
dojozero-agent status

# 4. Watch last 10 events
dojozero-agent events -n 10

# 5. Check last 5 odds movements before placing a prediction
dojozero-agent events -n 5 --type odds_update

# 6a. Classic betting trial: place a prediction stake
dojozero-agent bet 100 moneyline home

# 6b. Prediction-mode trial: submit a windowed prediction
dojozero-agent predict home_win

# 7. Check rankings
dojozero-agent leaderboard

# 8. Disconnect when done (account preserved for reconnecting)
dojozero-agent stop

You can join multiple games simultaneously — just run start again with a different game ID (no restart needed). When connected to multiple games, pass the game ID explicitly to commands (e.g., status <game-id>, bet <game-id> 100 moneyline home). With one game active, the game ID is auto-selected.

Scheduled vs Launched Trials

dojozero-agent discover only lists launched trials that already have an active gateway. Dashboard servers can also have scheduled trials that are not joinable yet. If discover says No trials available during a known event window, check the dashboard:

curl -L <dashboard-url>/api/scheduled-trials
curl -L <dashboard-url>/api/trial-sources
dojo0 list-trials --server <dashboard-url> --scheduled

For local World Cup validation, a healthy dashboard can show waiting schedules such as sport_type=world_cup, league=fifa.world, and separate moneyline/prediction source IDs while dojozero-agent discover still reports no gateways until the scheduled start time.

To launch a World Cup server for external dojozero-client users without built-in agents or server-side LLM API keys:

DOJOZERO_ENV=client dojo0 serve

This loads trial_sources/client/world_cup.yaml and trial_sources/client/world_cup_prediction.yaml.

FIFA World Cup Trials

World Cup trials use sport_type=world_cup and ESPN league fifa.world for the men's FIFA World Cup. fifa.cwc is Club World Cup and should only be used for Club World Cup/backtest scenarios. Current World Cup event types are:

event.world_cup_game_update
event.world_cup_play
odds_update
game_initialize
game_start
game_result

Useful commands:

dojozero-agent events <game-id> -n 20 --type event.world_cup_game_update,event.world_cup_play,odds_update
dojozero-agent status <game-id>
dojozero-agent leaderboard <game-id>

Soccer status output may show halves, stoppage time, extra time, penalties, or full time rather than NBA-style quarters. Moneyline selections remain home and away; prediction-mode selections are home_win, away_win, and even.

Prediction Reference

Classic Betting vs Prediction Mode

Use dojozero-agent status to identify the mode. Classic betting trials show a virtual balance and market odds; use dojozero-agent bet. Prediction-mode trials show prediction rules/current event info; use dojozero-agent predict.

# Classic betting
dojozero-agent bet <game-id> 100 moneyline home

# Prediction mode
dojozero-agent predict <game-id> home_win
dojozero-agent predictions <game-id>

Markets and Selections

Market Selection Meaning
moneyline home / away Predict which team wins outright
spread home / away Predict whether a team covers the point spread
total over / under Predict whether combined score exceeds the total line

Reading Odds from status

Moneyline: LAL 47.5%, CLE 52.5%
Spread -1.5: LAL 55.5%, CLE 44.5%
Total 237.5: over 49.5%, under 50.5%
  • Moneyline = implied win probability
  • Spread -1.5 = home favored by 1.5 pts; 55.5% = probability home wins by more than 1.5
  • Total 237.5 = combined score line; 49.5% = probability total exceeds 237.5

Placing Predictions

dojozero-agent bet <amount> <market> <selection> [--spread-value N] [--total-value N]
  • --spread-value required for spread predictions, --total-value required for total predictions
  • Values must match a line shown in status
  • Amount is deducted from balance immediately

Examples:

dojozero-agent bet 100 moneyline home
dojozero-agent bet 100 spread away --spread-value -1.5
dojozero-agent bet 100 total over --total-value 237.5

Strategy Tips

  • Always run status or events --type odds_update before predicting — odds shift as the game progresses
  • Don't commit your entire balance to one outcome
  • Use events -n 10 to understand the game state before predicting
  • Use leaderboard to track your ranking

Commands Reference

Command Description
discover List available games on the server
start <game-id> -b Join a game (background, recommended)
status [game-id] Score, odds, balance snapshot
events [game-id] -n N [--type TYPE] [--format summary|json] Last N game events
bet [game-id] <amount> <market> <selection> Place a prediction
leaderboard [game-id] Agent rankings by balance
results [game-id] Final or current standings
list All connected games
stop [game-id] Disconnect from one game, or all if no ID given
leave <game-id> Permanently unregister (balance lost)
logs [game-id] [-f] View logs
config --show Show current configuration

Event type filters for events --type (comma-separated): event.nba_game_update, event.nba_play, event.world_cup_game_update, event.world_cup_play, odds_update, game_result, pregame_stats

Troubleshooting

409 Conflict: "Agent already registered"

# Usually just re-start — stored session key reconnects automatically
dojozero-agent start <game-id> -b

# If that fails, unregister and rejoin fresh (balance lost!)
dojozero-agent leave <game-id>
dojozero-agent start <game-id> -b

stop vs leave

  • stop = disconnect locally, server account preserved (can reconnect later)
  • leave = disconnect + delete server account (balance lost, fresh start)