Research & Data

MrCha Skills — portable Agent Skills for ChatGPT, Codex, and compatible agents

Codex skill for organizing qualitative interview recordings into Feishu/Lark Minutes transcripts and structured research documents.

An extensible toolkit for automated evaluation of LLMs and AI agents, including rubric judges, datasets, scoring scripts, and regression tests.

Claude skills for automated literature surveys: verified corpora, honest metrics, minimal figures, and a short PDF report of the state of the art.

个人自律记账 / 资金池自我治理引擎：确定性 Python 规则引擎（公式 F0~F8 + 审批判定 §4.4）+ LLM 薄路由。含支取审批、冷静期、负债/房贷建模、第三方资产导入、审计留痕。Claw 搭档的 self-trust skill 本体。

Source-aware web search and readable page extraction for AI agents

本地优先、来源可追溯的个人投资研究助手，使用 Agent Plan、DataPro 与豆包搜索。

Personal Claude Code and Codex skills, including the idea research pipeline and XJU DOCX tooling.

从单客户原始沟通还原连续决策链并生成可验证销售案例的 Codex Skill

Agent skills for market analysis and market-moving signal watching (fxempire, hormuz-ais, truthsocial).

A repository for deploying agents/skills for parameter tuning, profiling and analysis.

A benchmark-informed Codex skill that uses Sol to plan, Terra to build, and Luna to handle mechanical work through evidence-gated handoffs.

Codex skill for Tony-style US stock daily analysis

This Codex skill combines Firecrawl web acquisition with a persistent, auditable research corpus.

Manage cannabis cultivation data to generate crop plans, compliance records, and operational workups.

Guide Codex through structured engineering tasks to manage complex AI projects, research, and multi-file code development.

A secure, self-hosted registry for managing and sharing AI Agent skills within your organization. Centralize workflows, enforce governance, and prevent data leaks. Features RBAC, version control, and seamless integration with your existing AI frameworks

自动检索、核验、去重并生成可执行外贸情报日报的 Codex Skill。A Codex Skill for researching, verifying, deduplicating, and generating actionable daily foreign-trade intelligence.

Convert PDF notes into cited answers, adaptive quizzes, and flashcards using local LLMs for private study.

Agent Skills

Agent Skill: give an AI agent a top-down US stock market read — regime verdict (risk-on or stand down), ranked leaders, and per-name opportunity/entry/hold scores for ~560 names. Free, no key, no wallet, no account.

Local-first, explainable US equity crowding-risk monitor | 本地优先的美股拥挤风险雷达

Diagnose why a GitHub Actions run failed and find its root cause; decide whether to retry that failed Action once; check or rank GitHub bounties; audit AGENTS.md readiness; detect MCP schema drift.

Wenyu privacy-aware Bazi analysis engine and Codex skill