Bilibili Content Creator

Agent的核心公式为：Agent = 大模型 + 工具调用 + 自主决策能力

The author argues that AI Agent learning can be broken down into four stages, progressing from zero to a practical developer within three months.

The Bilibili creator says that currently more than 85% of Vibe Coding-related videos are a mixed bag with inferior content passing as good, and that very few truly take learners through hands-on practice step by step and break down the core logic.

The creator provided a complete one-click installation script for the development environment; after receiving the course materials, double-clicking the script completes the full assembly of the basic development environment.

The creator stated that the LMWiki project, from setting up the development environment to final completion, involved a total of 10 rounds of professional prompts to DeepSeek Harness, and the development process was exceptionally stable and highly efficient, with his personal assessment surpassing the development experience of Codex and Cloud Code.

The video's author said that interested users can comment "missing invoice" to get this workflow template.

The AI Skill shared by the UP主 includes two independent functions, literature review and batch deep reading, aimed respectively at two different stages in the research workflow.

The Bilibili UP host suggests evaluating IT career direction choices from four dimensions: salary and benefits, job demand, capability requirements and future development.

The UP host mentioned a pragmatic goal — 'let us monetize the knowledge we have learned'.

Using traditional methods, some people may submit only dozens to over a hundred vulnerabilities per year, whereas with the help of AI-assisted vulnerability mining, some people have submitted more than 14,000 vulnerabilities in a year.

The content of this article originates from a single UP host's experience sharing, which contains marketing elements promoting their own courses.

The article has an obvious course-promotion nature, mentioning unverifiable personal experiences such as spending over 20,000 yuan on courses and landing a 30K job offer.

The UP host originally had a mechanical engineering background and later successfully transitioned into a large model position.

The UP host stated that they spent over 20,000 yuan on courses when switching careers, only to learn hollow theory.

In a DeepSeek Harness practical tutorial, a Bilibili creator demonstrated building an industrial-grade knowledge base LMWiki from scratch using only 10 rounds of prompts, 8 Skills, 1 Agent preset, and 2 feature plugins.

A Bilibili content creator used Doubao Work to hand 53 old drafts, cases, and topic selections from Feishu to AI, generating a searchable, filterable webpage that can jump back to the original text, all without writing any code, with the two rounds of tasks taking a total of 36 minutes and 22 seconds.

This deployment project is based on Q-engineering's open-source inference code, on top of which the uploader added three capabilities: web service, persistent model residency, and streaming output.

The Bilibili uploader, through the Devin Max plan, invoked Argon at its highest tier and used the same prompt to build six large applications, compiling, testing and reviewing the code one by one; the six applications had a total score of 60, and Argon got 32.6 points (about 54%).

The UP host spent more than 20,000 yuan on courses to learn AI Agent for a career change, but what was learned was empty theory, without knowing how to build workflows or implement scenarios, and during interviews could not answer practical details.

A Bilibili creator demonstrated a complete solution for automated sales lead distribution using n8n, GoHighLevel, and AI, with the workflow taking less than 1.5 seconds from form submission to notification being sent.