ai-agent-deep-dive
An in-depth research report on the source code of AI agents such as Claude Code, published as PDFs and mostly in Chinese. For readers who want to learn how these agents are built.
Share on XNo license declared
Overview
AI Agent Deep Dive is a research report on the source code of AI agents such as Claude Code, published as PDF files and written mostly in Chinese. The Claude Code report has a v2.1 edition with a chapter on the memory system. The repository also holds a small teaching agent in Python with an agent loop, a replaceable fake LLM, skills discovery and a CLI, meant for learning how an agent is structured.
Key features
- PDF research reports on agent source code
- Claude Code report v2.1 adds a memory system chapter
- Minimal teaching agent in Python with a CLI
- Fake LLM layer that can be swapped for a real model
Best for
Readers, mostly Chinese-speaking, who want to study how coding agents are built. The teaching agent does not call a real model API yet.
- Upstream
- tvytlx/ai-agent-deep-dive
- Fork on GitHub
- Guo-astro/ai-agent-deep-dive
- Upstream stars
- 5.8k
- Category
- Learning, interviews and curated lists
- Language
- Python
- License
- No license declaredWithout a license, the author keeps all rights. Ask the upstream owner before reusing the code.
- Forked
- 2026-04-01
- Sync status
- In syncLast synced 2026-10-10
More in Learning, interviews and curated lists
A Jupyter Notebook book on self-teaching for learning coding and study habits. It gives readers a practical path for learning by reading, practicing, and improving their own methods.
Forked 2026-10-08Last synced 2026-10-10No license declaredLearning, interviews and curated listsGitHub
A learning repository for Machine Learning Systems, bringing textbook concepts, labs, tools, and hardware context together so learners can study and build real AI engineering workflows.
Forked 2026-10-07Last synced 2026-10-10Custom license (see repository)Learning, interviews and curated listsGitHub
ThinkStats materials for the third edition, with notebooks, data, and supplemental files for learning statistics through examples and exercises in a runnable local setup.
Forked 2026-10-07Last synced 2026-10-10License: MITLearning, interviews and curated listsGitHub
A Spanish-language guide for software engineering technical interviews, published as a browsable website. It brings together topics and references for interview preparation and reviewing core engineering concepts.
Forked 2026-10-06Last synced 2026-10-10License: MITLearning, interviews and curated listsGitHub