Zhang Lu.
Turning complex judgment into a traceable pipeline, with Harness + Loop
C-suite decision agents · Designed for AI-native organizations
Building a few AI products that turn complex judgment into traceable pipelines: mbabrand (a jury of agents scoring brands), bossagent (a decision agent for leaders), and oaf.world (an AI investment-research workbench). I keep 30-some Claude Skills on my machine, and post notes on my WeChat account “Zhang Lu's Murmurs.”
Claude SkillsMulti-agentBrand judgmentDecision agentsAI investing research
A suite of C-suite decision agents
About the suite →Designed for AI-native organizations: turning the CEO's strategy calls, the CMO's brand calls, and the CFO's investment-research calls each into a pipeline you can research, score, and review.
Featured projects
All → 
Second Brain
2026A compounding personal Wiki knowledge base continuously written and maintained by an LLM—knowledge is compiled once, then kept fresh forever (a Wiki is not RAG)
Claude CodeMarkdownWikiCloudflare Pages
56 roadshow projects, from on-site photos to a publishable DD panorama page, all through one pipeline
Claude SkillMulti-agentCloudflare Pages
An unofficial recap site turning the 30 talks from the 2026 AI Vision Forum (Shanghai) into minutes, a knowledge base, and a knowledge graph
HTMLJavaScriptKnowledge graphCloudflare Pages ≈ 6,376 LOC
SiliconForge
2026Built for business requesters
Requirements in, software out
CloudflareMachine-readable PRDParallel AI squadsExecutor verification beta
excel-ai-analyst
2026Upload a spreadsheet, see through its structure first
SheetJSIn-browser computeZero uploadFormula-chain parsing
Turn one short conversation into a story your work can tell
ChatGPT VoiceRealtime voiceAuto transcriptionSix-dimension unified scoring
From experimental psychology toward brain science × AI
Opportunity aggregationAdvisor radarPaper trackingStatic snapshot
FreeModel Port
2026Built for AI developers
A discovery, live-verification, and local access layer for free LLMs worldwide, built to connect Codex and other AI CLIs safely
PythonFastAPIModel RoutingLocal-first
Hand the AI the outcome, not just the answer
Tauri 2ReactFastAPIMCP beta
Decks
All →Turn brand influence into an intelligent asset you can monitor, score, compare, and review. 7 dimensions researched in parallel × 43-judge deliberation × 5-lens scoring, with sentiment-driven versioned re-audits.
Web deck mbabrand.com
Hand your important calls to a panel deliberation: 1 anchor conviction + N dimensional judges scoring independently, mandatory opposing view, producing judgment reports that are citable, scorable, and falsifiable.
Web deck bossagent.cc
A quick-look deck for a satellite-internet decision agent plus an AI research workbench spanning the US, A-share, and Hong Kong markets—numbers come from tools, narrative from the LLM, gaps left blank rather than fabricated.
Web deck oaf.world
Zhifang · OpenAgent
2026Incubating AI super-individuals.
Site openagent.ltd
Recent writing
All → · On-site
Localisation isn't translation: shipping a Chinese build of an agent runtime
The hard part wasn't the strings — before 0.1.7 one auto-update turned it back into English. Plus two designs worth stealing from OpenWorker's source: concurrent reads with sequential writes, and four permission levels.
#agent#localisation#source analysis#desktop
· On-site
A list of free models is the wrong abstraction
A list records claims, not evidence. FreeModel Port splits candidates from runnable models, with one real request drawing the line — plus why a unified router is inherently a high-privilege centre.
#LLM#architecture#local-first#trade-offs
· Blog
Free LLMs should not depend on luck
A feature story on FreeModel Port: continuously discovering free LLMs, verifying them with real requests, and connecting them safely to Codex and AI CLIs through a local endpoint.
#LLM#AI CLI#free models#local-first
· On-site
Making the people who do not present well visible
Achievement reviews fail two ways: the smooth talker wins, and plans get counted as results. This method removes both with a 1–2 minute voice interview, adaptive follow-ups and one scoring ruler.
#AI review#voice#achievement scoring#method
· On-site
The opportunities are not hard to find — they are scattered and time-boxed
A brain-science × AI opportunity radar for the class of 2027. The scarce thing is not a list of opportunities but the ordering — what you have to do first for the next thing to be possible.
#brain science#job hunting#aggregation#radar
Skills
All 42 →A browser-automation CLI for AI agents. Use it whenever you need to open a page, fill a form, click a button, take a screenshot, scrape data, test a web app, or automate any browser task — in preference to any built-in browser automation or web tool.
Build AI agents on Cloudflare Workers with the Agents SDK. Load it when you're building stateful agents, durable workflows, real-time WebSocket apps, scheduled tasks, MCP servers, chat / voice agents, or browser automation.
Send and receive transactional email with Cloudflare Email Service (Email Sending + Email Routing). Use it when integrating email in Workers / Node / Python / Go, or handling SPF / DKIM / DMARC deliverability — even when it looks simple, there are configuration details that matter here.
Full-stack Cloudflare platform skill. Covers any Cloudflare development task: Workers / Pages / storage (KV, D1, R2) / AI (Workers AI, Vectorize, Agents SDK) / feature flags / networking / security / IaC (Terraform, Pulumi).
An end-to-end roadshow / demo day dossier pipeline. Project card screenshots + booth-wall photos + official articles → structured JSON + panoramic HTML + a DD table (7-way parallel research) + a Word report + CSV + one-click deploy to Cloudflare Pages.
Create and review Cloudflare Durable Objects. Use for stateful coordination (chat rooms, multiplayer games, booking systems), RPC, SQLite storage, alarms, WebSockets, or reviewing DO code against best practices.
plus 27 more skills: /boss , /ego-browser , /lark-approval , /lark-apps , /lark-base , /lark-attendance , /lark-calendar , /lark-doc , /lark-contact , /lark-event , /lark-im , /lark-drive , /lark-mail , /lark-markdown , /lark-minutes , /lark-openapi-explorer , /lark-skill-maker , /lark-note , /lark-shared , /lark-slides …
How this site works
See the architecture →Content is plain text in git; one command compiles it into a full set of static artifacts; pushing deploys it — no CMS, no database, no server. The same file becomes pages, JSON endpoints, RSS and the search corpus at once.
markdown → Zod check → Astro build → pages · JSON · RSS → CF Pages
Past work
Details →Before these AI products —— tui3.com rebuilds several Chinese personal sites I worked on between 1999–2007, from server archives into static sites, preserved for good.
