rapid-expert

Rapid Expert (Chinese: Ji Su Zhuan Jia) is an interactive command-line agent designed to help someone with no background in a field reach practical competence in 5 to 12 days. The primary use cases described are investment research, startup validation, job hunting, consulting analysis, and product development. Instead of delivering a static collection of documents or a general chatbot, it guides the user through a structured learning path, retrieves current materials from the web and GitHub, and produces real deliverables like industry analysis reports or competitor comparisons.

The project is written in Python and runs locally by default, calling external AI model APIs when needed. On first launch it asks the user to configure which AI provider and model to use. After that, the main entry point is a command called study hacker that opens an interactive shell. From there, the agent interviews the user about their domain, current knowledge level, daily time budget, and risk boundaries, then generates a customized learning plan and starts executing it step by step.

Under the hood there is a harness layer that breaks work into fixed stages: risk assessment, planning, source scanning, ranking, GitHub search, building, and evaluation. Each stage can be run in sequence, paused for human review, resumed, or approved through a permission gate before sensitive actions (like network searches) proceed. The harness also exposes an HTTP API and an MCP (JSON-RPC over stdin/stdout) interface so external agent platforms can call it programmatically. A file-based task queue allows multiple jobs to run in sequence with a worker process.

A lightweight retrieval system lets the agent index sources and search them by query during a session. All runtime files including session logs, outputs, and configuration are kept local and excluded from version control.

The README is written primarily in Chinese. The project carries an MIT license and includes an explicit disclaimer that it does not provide legal, investment, medical, or other professional advice. High-stakes decisions in regulated fields must be confirmed by qualified professionals.