I am the Technical Staff of Myself.

I build tools for AI agents, from knowledge visualization and modular skills to agent packaging and security infrastructure.

I work at the intersection of AI research, agent infrastructure, and hands-on tooling.

I map the AI landscape by turning 263 industry reports into an interactive knowledge graph with nearly 4,000 nodes. I build infrastructure that makes agents portable and operational, from environment packaging to a growing set of modular skills covering search, scraping, memory, security, and media workflows. And I dig into how leading agent systems work under the hood, contributing to OpenClaw and studying projects like OpenCLI, Hermes Agent, and Claude Code's architecture.

Email: luosongred@gmail.com

Stack: Python / TypeScript / JavaScript / Shell / HTML/CSS

• OpenClaw: upstream contribution workspace focused on CLI/DX and agent runtime details.
• AgentBox: portable packaging for agent settings, skills, memory, and session context.
• AI Knowledge Graph: interactive AI ecosystem map built from 263 reports. Demo
• AI Personality Test: a deployable static quiz with Cloudflare Pages monitoring. Site
• AI Info Radar: media discovery and digest workflows across 50+ AI sources.

I actively work through upstream agent codebases instead of only reading about them.

• openclaw: recent work centers on cron UX, help output, and clearer memory-search failures.
• opencli: universal CLI hub for websites, Electron apps, and local tools.
• hermes-agent: self-improving agent framework from Nous Research that I track and study.
• claude-code-source-code: source snapshot kept for architecture research and study.

• agentbox: portable export/import for agent environments.
• openclaw: upstream contribution workspace for the OpenClaw personal AI assistant.
• opencli: universal CLI hub for websites, Electron apps, and local tools.
• hermes-agent: self-improving agent framework that I track and study.
• claude-code-source-code: source snapshot for architecture research and study.

• ai-knowledge-graph: interactive knowledge graph of the AI ecosystem.
• ai-info-radar: AI news and media radar across 50+ sources.
• ai-talent-graph: academic researcher profiling with OpenAlex, arXiv, Semantic Scholar, and ORCID.
• ai-talent-radar: recruitment-oriented talent search combining research, GitHub, and Chinese platforms.
• agent-job-monitor: daily monitoring of agent and LLM job openings across major AI companies.
• ai-personality-test: static AI personality quiz with monitoring and analytics.

• all-net-search-read: search and distill content across WeChat, Xiaohongshu, X, YouTube, Reddit, Bilibili, and Weibo.
• web-access: browser automation, search, and direct HTTP access for agents.
• wechat-reader: convert WeChat Official Account articles to Markdown.
• x-twitter-scraper: public X/Twitter scraping for profiles, timelines, and tweets.
• weibo-scraper: public Weibo extraction without login.
• bilibili-video: Bilibili download and local audio transcription.
• xiaoyuzhou-podcast: Xiaoyuzhou podcast fetch, transcription, and summarization.
• book-hunter: ebook discovery across Z-Library and Anna's Archive.

• persistent-memory: 3-layer memory distillation for long-term agent memory.
• agent-self-audit: 8-point health check for OpenClaw agents.
• lobster-guard: identity and prompt-injection guardrails for group-chat agents.
• skill-security-guard: skill code security scanner with A-F ratings.
• skill-design-guide: methodology and linting for building production-grade skills.

• NanoBanana-PPT-Skills: AI-generated PPT visuals and transition videos.
• Youtube-clipper-skill: semantic YouTube clipping, subtitle translation, and repackaging.
• guizang-s-prompt: prompt library spanning image, text, and video generation.

This account is where I build, test, and document how AI agents can search, read, remember, operate tools, and package their own working context.