Categories
Every choice in life deserves to be documented
Housing Reviews
Rent, noise, lighting, renovation, landlord rating... Find your ideal home without surprises
Learn More
Restaurant Reviews
Cuisine, price, taste, ambiance, must-try dishes... Never have a bad meal again
Learn More
Job Interviews
Company, position, difficulty, salary, interview process... Level the playing field
Learn MoreReal Data
Structured reviews, not "pretty good lol"
Every review has precise dimensions and ratings, enabling AI Agents to query like a database
East Village 1BR, quiet but pricey
East Village, New York
$2,500/mo · Noise 2/5 · Landlord 4/5
Bashu Hotpot, authentic Sichuan
Flushing, New York
$12/person · Taste 5/5 · Ambiance 3/5
Google L4 SWE interview, got offer
Mountain View, CA
Difficulty 4/5 · 3 months prep · TC $280K
Why OpenBook
Information equality starts with structured data
Structured Data, Not Noise
Every review is defined by a Schema with typed fields. Not forum posts, but precisely queryable structured data.
Agent-Native, Human-Friendly
Just talk. Your AI Agent reads the Schema, collects info, validates data, and submits the review.
Non-Commercial, Forever Free
No ads, no paywalls, no recommendation algorithms. Data belongs to the community, not corporations.
Everyone Can Build Their Own
One-click Fork from GitHub Template, customize the Schema, and you have your own review library.
Comparison
OpenBook vs Traditional Platforms
| OpenBook | Xiaohongshu/Yelp | BBS/Forums | |
|---|---|---|---|
| Data Structure | Schema-enforced structured | Mostly free text | Plain text |
| Agent Readable | Native support | Requires scraping | Requires scraping |
| Precise Query | Field-level filtering | Keyword fuzzy search | Full-text search |
| Commercialization | Free & open forever | Ads + paid promotion | Varies |
| Data Ownership | Community (GitHub) | Platform-owned | Platform-owned |
| Replicability | One-click Fork | Not replicable | Requires setup |
How It Works
Chat to publish, ask to search
Connect your AI Agent via MCP or Skills, and do everything through conversation
Tell your Agent what you want
"I want to review my apartment" or "Find me a quiet 1BR in NYC"
Agent reads Schema, guides the conversation
"Monthly rent?" → "Noise level? 1-5" → "Renovation condition?"
Structured data saved automatically
Agent validates format, generates JSON, commits to GitHub repo
Get Started
Three ways to connect to OpenBook
MCP Server
Use directly in MCP-compatible clients like Claude Desktop, Cursor, etc.
Agent Skills
Use as a Skill in Agent platforms like Manus, OpenClaw, etc.
Schema-Driven
One YAML file defines everything
Schema tells the Agent what data to collect, what format, and how to ask the user
name: housing
display_name: "Housing Reviews"
fields:
rent_monthly:
type: number
required: true
unit: "USD"
agent_prompt: "What's the monthly rent?"
noise_level:
type: number
required: true
min: 1
max: 5
agent_prompt: "How's the noise? 1=very quiet, 5=very noisy"
renovation:
type: string
required: true
enum: [new, good, average, old, poor]
agent_prompt: "How's the renovation condition?"