Ryanyx
中文
← All projects
AI & Automation / PROJECT NOTES

Local ReAct Automation Agent

A learning prototype that uses a tool loop for file search, note writing, and webpage summaries.

PrototypeFirst located record · 2026-06Archive checked · 2026-10-26
PythonLangChainReActTelegram
01 / THE INTENT

Make the observe, reason, act, and feedback loop concrete.

02 / THE FLOW

How it works

  1. 01Send a task
  2. 02Observe context
  3. 03Call file/web tools
  4. 04Return feedback
03 / THE WORK

What exists

01

CLI, web, and HTTP entry points

02

Telegram and WhatsApp adapters

03

Scoped workspace and a limited tool set

04 / CURRENT POSITION

Where the project stands

Local implementation and teaching material exist. An independent learning project, not an official OpenClaw project.

CONNECTED WORK

Nearby projects

DeepDesk · Local Agent DeskA chat workspace for model conversations, tool calls, action approval, and execution feedback.Forum Corpus & Local RetrievalA traceable local corpus pipeline for paginated threads, incremental updates, topic browsing, and compact exports.What Just Happened?Turns an AI coding task into a readable micro-lesson: what changed, why it works, and how it was checked.

AI-assisted English translation of the Chinese project record.