MoltMarkets Trading Agent

Complete MoltMarkets trading agent setup with autonomous trader, market creator, and resolution crons. Use when setting up a new MoltMarkets agent, configuring trading bots, or replicating the bicep agent architecture. Includes Kelly criterion betting, learning loops, and degenerate trader personality.

This skill implements an autonomous trading system that will use your MoltMarkets API key to place bets, create markets, and resolve outcomes. Before installing: (1) review scripts/setup.js and all referenced files to see exactly what network calls and file reads/writes they perform; (2) don't use your main/real-money account — create a limited-scope or test API key if possible; (3) note that the skill assumes Node is available and cron infrastructure is present, but these are not declared — verify and correct paths before running; (4) consider disabling autonomous invocation (set disableModelInvocation: true) or require explicit user approval for any trade/resolution actions; (5) inspect and possibly restrict the credentials file (permissions) and confirm the skill won't exfiltrate keys to unexpected endpoints (Binance/HN Algolia are mentioned as oracles, but verify exact URLs used). If you want help reviewing the setup.js contents or converting run instructions to a safer, explicit checklist, provide the script and cron definitions and I can examine them.

Type: OpenClaw Skill Name: Developer: Version: Description: OpenClaw Agent Skill Suspicious High-Entropy/Eval files: 1 The skill bundle is classified as suspicious due to the explicit instructions for the AI agent to execute direct shell commands (e.g., `curl`, `jq`, `date` command substitution) within its prompt messages (`payload.message` in `references/cron-definitions.md`). While these commands are currently used for legitimate interactions with the MoltMarkets API (api.zcombinator.io) and public oracle APIs, the mechanism of instructing an AI agent to execute arbitrary shell commands is a high-risk capability that could be exploited via prompt injection. Additionally, the skill instructs the creation of persistent cron jobs for the agent's operation, and includes instructions for the agent to operate 'SILENTLY' or reply 'NO_REPLY', which, while potentially for verbosity control, also reduces visibility of the agent's actions.