← 返回 Skills 市场
Mi Analista
作者
davidpolotm
· GitHub ↗
· v1.0.0
540
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install mi-analista
功能描述
Trade 10-second crypto prediction markets on PredictMe
安全使用建议
This skill will register an agent on https://api.predictme.me, obtain a one-time API key, and save that key and agentId to a local file (~/.predictme/credentials.json) so it can poll /odds and place bets automatically. The package metadata did not declare any required credentials or config paths — that mismatch is a red flag. Before installing: (1) verify the PredictMe site and the skill author (unknown source/homepage), (2) be comfortable with the agent storing an API key on your machine and check/limit file permissions or prefer encrypted storage, (3) consider creating a throwaway account (TEST balance only) rather than using any account tied to real funds, (4) enable any 'requireApproval'/manual-approval preference so the agent cannot bet without your explicit go-ahead, and (5) review the skill files (skill.md and sw.js) yourself or have a developer inspect how credentials are stored and used. If you can't verify the publisher or are uncomfortable with automatic credential storage and autonomous betting, do not install.
功能分析
Package: ClawPredict - Grab the Future (xpi)
Version:
Description: 10-second crypto predictions. Grab the future. Predict. Win.
The package consists primarily of static assets (SVG images, HTML, manifest files, robots.txt, sitemap.xml) and documentation files (JSON, Markdown) describing an API for AI agents to interact with a crypto prediction market. The `sw.js` service worker implements standard caching strategies and explicitly avoids intercepting API requests or external network calls. There is no executable code that performs any malicious actions such as data exfiltration, unauthorized access, content injection, or cryptocurrency mining. The package appears to be a legitimate Progressive Web App (PWA) or browser extension for a crypto prediction platform, with a focus on AI agent interaction for simulated trading.
能力评估
Purpose & Capability
SKILL.md describes an AI agent that registers with PredictMe, receives a one-time API key, polls market endpoints, and places bets — behavior coherent with the stated purpose. However the package metadata declares no required credentials or config paths while the instructions explicitly require storing an apiKey/agentId and a nonce in ~/.predictme/credentials.json (and keeping preferences.json). The lack of declared credentials/config paths is an incoherence.
Instruction Scope
Runtime instructions direct the agent to (1) prompt the owner for an email, (2) register the agent, (3) poll for admin approval and retrieve a ONE-TIME apiKey, and (4) save that apiKey and agentId to a local file (~/.predictme/credentials.json). They also instruct ongoing polling of /odds and automated betting logic. These steps access and persist sensitive secrets and perform network operations. The instructions do not request the owner to paste the key but instead instruct the agent to store it automatically — this increases the risk surface and is not reflected in declared requirements.
Install Mechanism
No install spec is provided (instruction-only), so nothing arbitrary is downloaded or installed. A small service worker (sw.js) is included; it performs benign caching of local static assets and does not appear to introduce remote code fetches. Low install risk.
Credentials
No environment variables or primary credential are declared in the registry metadata, yet SKILL.md requires a persistent agent apiKey (pm_agent_*) and saving it to a local file. Requesting persistent agent credentials is proportionate to a trading skill, but the absence of declared credentials/config paths and lack of guidance about secure storage (file permissions, encryption) is a notable mismatch and security oversight.
Persistence & Privilege
The skill does not request always:true and uses standard autonomous invocation (disable-model-invocation is false). It instructs the agent to persist credentials locally so it can operate over time; autonomous invocation combined with stored credentials means the agent could place bets without owner interaction. That autonomous capability is expected for this class of skill but increases risk if the source is untrusted.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install mi-analista - 安装完成后,直接呼叫该 Skill 的名称或使用
/mi-analista触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
predictme skill 1.3.0 changelog
- New skill name: changed from "mi-analista" to "predictme"
- Updated description: now focused on trading 10-second crypto prediction markets
- Complete rewrite of skill documentation, including:
- Step-by-step agent registration and credential management
- Detailed onboarding (observation, paper trading, strategy scaling)
- In-depth explanation of market mechanics, grid betting, and decision rules
- Sample code and JSON for preferences and operational flow
- Added guidance for safe, disciplined trading (risk management, bet sizing, scheduling)
元数据
常见问题
Mi Analista 是什么?
Trade 10-second crypto prediction markets on PredictMe. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 540 次。
如何安装 Mi Analista?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install mi-analista」即可一键安装,无需额外配置。
Mi Analista 是免费的吗?
是的,Mi Analista 完全免费(开源免费),可自由下载、安装和使用。
Mi Analista 支持哪些平台?
Mi Analista 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 Mi Analista?
由 davidpolotm(@davidpolotm)开发并维护,当前版本 v1.0.0。
推荐 Skills