What is this?
This is a book (Zhijian AI Bluebook ZJBB-003) that reverse-engineers WorkBuddy, an AI agent app, tracing its UI behavior back to the local files it writes and the requests it sends to the model, covering prompts, memory, plugins, experts, Skills and security boundaries.
This is a book (Zhijian AI Bluebook ZJBB-003) that reverse-engineers WorkBuddy, an AI agent app, tracing its UI behavior back to the local files it writes and the requests it sends to the model, covering prompts, memory, plugins, experts, Skills and security boundaries.Is this software I install, or something I read?It is a self-contained ebook, not an application. You read it online via GitHub Pages, as a PDF, or as a single offline HTML file.
It is a self-contained ebook, not an application. You read it online via GitHub Pages, as a PDF, or as a single offline HTML file.How do I read it?Open the GitHub Pages online version in a browser, download book.pdf for A4 reading, or grab the self-contained book.html for offline use; all are also attached to the v0.1.0-pilot Release.
Open the GitHub Pages online version in a browser, download book.pdf for A4 reading, or grab the self-contained book.html for offline use; all are also attached to the v0.1.0-pilot Release.Is this an official WorkBuddy manual, or will it go stale?The analysis is pinned to a specific WorkBuddy build observed locally on macOS and is independent research with no affiliation to WorkBuddy's maker, so paths and defaults may drift as WorkBuddy updates.
The analysis is pinned to a specific WorkBuddy build observed locally on macOS and is independent research with no affiliation to WorkBuddy's maker, so paths and defaults may drift as WorkBuddy updates.Who is this book for?People who use WorkBuddy daily and hit unexplained behavior, Agent practitioners who want to understand how prompts/memory/plugins combine into a request, and engineers or trainers who need to teach Harness mechanics, especially those already familiar with Claude Code or Codex.
People who use WorkBuddy daily and hit unexplained behavior, Agent practitioners who want to understand how prompts/memory/plugins combine into a request, and engineers or trainers who need to teach Harness mechanics, especially those already familiar with Claude Code or Codex.How is the book organized?It has 15 chapters in 5 parts plus 4 appendices: from UI behavior to local files, how a request is assembled, how capabilities (experts, plugins, Skills, Connectors, MCP) enter the main thread, security/ecosystem comparisons with Claude Code and Codex, and a practical troubleshooting workflow.
It has 15 chapters in 5 parts plus 4 appendices: from UI behavior to local files, how a request is assembled, how capabilities (experts, plugins, Skills, Connectors, MCP) enter the main thread, security/ecosystem comparisons with Claude Code and Codex, and a practical troubleshooting workflow.Does it cover MCP and Connectors?Yes, part three (chapters 7-11) explains experts, plugins, Skills, session evidence, Connectors and MCP, and how they inject capabilities into WorkBuddy's main thread.
Yes, part three (chapters 7-11) explains experts, plugins, Skills, session evidence, Connectors and MCP, and how they inject capabilities into WorkBuddy's main thread.What exact WorkBuddy version does the research target?The bluebook's findings are pinned to WorkBuddy 5.3.13 (build 20fd9da5c9a0cac41feb61b133f7bce4fe183a89), observed locally on macOS, with research cut off on 2026-08-15 and backed by 12 sources and 27 claims.
The bluebook's findings are pinned to WorkBuddy 5.3.13 (build 20fd9da5c9a0cac41feb61b133f7bce4fe183a89), observed locally on macOS, with research cut off on 2026-08-15 and backed by 12 sources and 27 claims.Where should I start if I just want WorkBuddy to work reliably?The recommended quick path is chapters 1 to 3, 5, 7 and 12, then 14 for the five-step troubleshooting method; deeper customization or teaching adds chapters 4, 6, 8-10, 13 and 15.
The recommended quick path is chapters 1 to 3, 5, 7 and 12, then 14 for the five-step troubleshooting method; deeper customization or teaching adds chapters 4, 6, 8-10, 13 and 15.What are similar projects?ai-agent-book, harness-engineering, claude-code-book, hermes-agent-orange-book, ai-system-design, vibe-coding-cn, operator-memory, lieflat-gongwen
ai-agent-book, harness-engineering, claude-code-book, hermes-agent-orange-book, ai-system-design, vibe-coding-cn, operator-memory, lieflat-gongwenWhich topics is it in?workbuddy, harness-engineering, chinese-language, ebook, context-engineering, agent-skill, harness, claude-skills, agent-memory, knowledge-base, prompt-engineering, book
workbuddy, harness-engineering, chinese-language, ebook, context-engineering, agent-skill, harness, claude-skills, agent-memory, knowledge-base, prompt-engineering, book