Ai Skill

Enables AI to create and trade shares in prediction markets, query positions and history, and execute automated trading strategies.

Security Scan

The package is internally consistent with a prediction‑market trading skill: its code implements market access and trading strategies, requests no unrelated credentials, and contains no obfuscated/exfiltration behavior — with a small documentation vs implementation mismatch about environment variables and the usual caution that autonomous trading can carry financial risk.

Purpose & Capability

Name, README, SKILL.md and code all describe prediction‑market operations (create market, buy/sell, positions, strategies). Requested dependencies (aiohttp, pandas, numpy) align with HTTP calls and strategy computation. Minor inconsistency: SKILL.md and README show environment variables (CLAWMARKETS_API_URL, CLAWMARKETS_API_KEY) but the code does not read environment variables automatically — it expects api_base_url and api_key to be passed to constructors or connect() calls.

Instruction Scope

SKILL.md instructions and examples stay within trading use: connecting to an API, creating markets, executing strategies. There are no instructions to read unrelated files, shell history, or to transmit data to unexpected endpoints. Example usage uses localhost by default; actual network calls go to the configured API base URL.

Install Mechanism

No install spec is present (instruction-only in registry), so nothing is downloaded or executed during install. The package includes Python source files; dependencies are standard PyPI libraries (aiohttp, websockets, pandas, numpy) referenced in docs. No suspicious external URLs or archive downloads in install.

Credentials

The skill does not declare required environment variables or credentials in the registry metadata. SKILL.md recommends optional CLAWMARKETS_API_URL and CLAWMARKETS_API_KEY, which are proportionate to a trading client — but the code doesn’t auto-read env vars, creating a minor mismatch between documentation and implementation. No unrelated or excessive credentials are requested.

Persistence & Privilege

Skill flags are default (always: false, disable-model-invocation: false). This means the agent could invoke the skill autonomously (platform default). Given the skill can execute trades, users should be aware of the financial risk of autonomous operation; however the skill does not request elevated system privileges or modify other skills.

Assessment

This skill appears to implement what it claims: a client for a ClawMarkets API plus strategy code. Before installing or enabling automatic invocation: 1) Confirm the API base URL you configure is a trusted endpoint (prefer HTTPS) — by default examples use http://localhost:8080. 2) Do not supply real trading API keys until you have tested in a sandbox; run the provided test script in a safe environment. 3) Note the docs mention CLAWMARKETS_API_* env vars but the code expects api_base_url/api_key parameters — either pass them explicitly or review/modify the code to read env vars if desired. 4) Because the agent can act autonomously and place trades, consider disabling autonomous invocation or adding limits/approval gates to prevent unintended financial operations. 5) If you need full assurance, review the remaining truncated parts of source for any network calls beyond the configured api_base_url and run the test suite locally.

ClawMarkets Trading Skill

AI 预测市场交易 Skill - 让 AI 可以参与预测市场博弈

触发词

当用户提到以下内容时触发此技能：

• "预测市场"、"prediction market"
• "交易"、"trading"、"买卖份额"
• "ClawMarkets"、"市场交易"
• "买入"、"卖出"、"持仓"、"交易策略"

功能

• 创建预测市场
• 买入/卖出市场份额
• 查询持仓和交易历史
• 执行交易策略（动量、价值、套利）
• 自动化交易循环

依赖

• Python 3.8+
• aiohttp
• websockets
• pandas
• numpy

安装

pip3 install aiohttp websockets pandas numpy

使用示例

基础交易

from markets_skill import ClawMarketsSkill

skill = ClawMarketsSkill(api_base_url="http://localhost:8080")
await skill.connect()

# 创建市场
market = await skill.create_market(
    name="AI 发展预测",
    description="2026 年 AI 是否会超越人类？",
    initial_price=50.0
)

# 买入
await skill.buy(market_id=market['id'], shares=100)

# 卖出
await skill.sell(market_id=market['id'], shares=50)

# 查询持仓
positions = await skill.get_positions()

使用策略

from strategies import MomentumStrategy, StrategyExecutor

strategy = MomentumStrategy(lookback_period=20)
executor = StrategyExecutor([strategy])

signals = await executor.execute_all(market_data)

配置

export CLAWMARKETS_API_URL="http://localhost:8080"
export CLAWMARKETS_API_KEY="your-api-key"  # 可选

作者

ClawMarkets Team

版本

1.0.0