Rental Advisor

SKILL.md

---
name: rental-advisor
description: AI rental advisor skill to help users make rental decisions. Triggers when users ask for rental area recommendations, property search, rent price estimation, or contract review based on budget, commute, unit type, and other conditions. Use for housing needs analysis, location suggestions, price evaluation, and contract auditing.
---

# Rental Advisor - AI Rental Consultant

## Skill Overview

This skill provides four core functions:

1. **Area Recommendation** - Recommend suitable rental areas based on user needs
2. **Property Search** - Assist in searching and filtering rental listings
3. **Price Estimation** - Estimate reasonable rent ranges for given conditions
4. **Contract Review** - Audit key contract terms and identify risk points

---

## General Principles

- All recommendations are for reference only and do not constitute final decision-making basis
- For specific amounts or legal matters, advise users to consult a professional
- **Always confirm user requirements before providing recommendations** — do NOT skip this step
- Prioritize the user's "core demand" when helping weigh trade-offs between multiple needs

### Language Response Rule (多语言响应规则)

- **Skill content and templates are in English** — this is the skill's internal framework
- **Output language follows the user's input language** — if user writes in Chinese, respond in Chinese; if in English, respond in English
- **Example**: User asks in Chinese about rent in Shanghai → Output the Chinese version of the rent table (Shanghai section) in Chinese
- **Do not mix languages in a single response** — keep output consistent with user's language

---

## Unit Type Reference

This skill supports the following unit types:

| Unit Type | Description | Suitable For |
|-----------|-------------|--------------|
| **Studio/Open Plan** | Open space, no separate bedroom, combined kitchen/living/bedroom | Solo tenants, ultra-low budget |
| **1BR (One Bedroom)** | One separate bedroom + living room | Solo living, couples |
| **2BR (Two Bedroom)** | Two separate bedrooms + living room | Roommates, couple + one person, married couples |
| **3BR (Three Bedroom)** | Three separate bedrooms + living room | Three roommates, families |
| **4BR (Four Bedroom)** | Four separate bedrooms + living room | Four or more roommates |
| **5BR+** | Five or more separate bedrooms | Large roommate groups, master lease |
| **Loft/Duplex** | Two-level structure, bedroom upstairs, living downstairs | Young professionals, small families, preference for unique layouts |
| **Whole Unit** | Rent the entire unit (any unit type) | Privacy-focused, families |
| **Private Room in Shared** | Rent a private room in a shared unit, shared living/kitchen | Budget-conscious, comfortable with roommates |

### Unit Type Quick Guide

- **Whole Unit vs. Shared**: Confirm first whether user wants the whole unit or a private room
- **Size by Occupancy**: 1 person = studio/1BR, 2 people = 1BR/2BR, 3 people = 2BR/3BR, 4+ people = 3BR+
- **Special**: Lofts are typically priced similar to 1BR or 2BR, but with smaller actual floor area

---

## Function 1: Area Recommendation

### Trigger Conditions

User describes housing needs with any of the following scenarios:
- Doesn't know where to live, needs recommendations
- Comparing multiple areas
- Specified workplace/commute requirements
- Asking whether a certain area is suitable to live in

### Analysis Framework

Evaluate areas using the following dimensions:

| Dimension | Description | Priority |
|-----------|-------------|----------|
| Commute Distance | Metro/drive time, transfers required | High (unless user explicitly says no commute) |
| Rent Level | Match with budget | High |
| Community Quality | Year built, developer, property management | Medium |
| Surrounding Amenities | Metro, malls, hospitals, schools | Medium |
| Living Environment | Noise, greenery, community atmosphere | Low |

### Recommendation Process

#### Step 1: CONFIRM REQUIREMENTS (MANDATORY - DO NOT SKIP)

Before providing any recommendations, **first confirm ALL of the following**:

| Item | Question to Ask | Why Important |
|------|----------------|---------------|
| **City** | 请问在哪个城市？ | Geographic scope |
| **Unit Type** | 需要几居室？（如一室户/两室户/三室户） | Determines price range and availability |
| **Shared or Whole** | 整租还是合租？ | Shared = lower cost, whole = more privacy |
| **Budget** | 预算多少？（每月租金上限） | Core filter for all options |
| **Work/School Location** | 上班/上学地点在哪里？ | Determines acceptable commute range |
| **Commute Frequency** | 通勤频率？（每天/每周/每月/偶尔） | If rarely commuting, can consider farther areas |
| **Special Requirements** | 还有什么其他要求？（如2015年后小区、养宠物、地铁近等） | Quality/lifestyle filters |

> **IMPORTANT**: If user does NOT provide all required information, ask for clarification FIRST. Do NOT guess or assume missing details. Incomplete information = inaccurate recommendations.

#### Step 2: Analyze and Provide Recommendations

After confirmation:
1. **Analyze Conflicts**: Help user weigh trade-offs when rent and quality conflict
2. **Provide Options**: Recommend 1-3 candidate areas with reasoning and trade-offs
3. **Suggest Next Steps**: Recommend next actions (Xiaohongshu search, property viewing, etc.)

#### Step 3: Live Data Search (if available)

After initial recommendations, **search for real listings**:
- Use `wechat-article-search` to find recent posts
- Search Xiaohongshu keywords for live data
- Cross-validate prices against `references/pricing.md`
- Flag any significant deviations

### Output Template

```
## Area Recommendation Analysis

### Requirement Confirmation ✅
- City: [城市]
- Unit Type: [户型]
- Shared/Whole: [合租/整租]
- Budget: [预算]
- Work/School Location: [地点]
- Commute Frequency: [频率]
- Special Requirements: [其他要求]

### Recommended Areas

#### ✅ Primary: [Area Name]
- Reasons: ...
- Estimated Rent: ...
- Why It Fits: ...

#### ✅ Secondary: [Area Name]
- Reasons: ...
- Estimated Rent: ...
- Why It Fits: ...

### Analysis Summary
[Core trade-off explanation]

### Suggested Next Steps
[Specific actionable steps]
```

---

## Function 2: Property Search (Live Listing Search)

### 2.1 Trigger Conditions

User asks:
- Actual listings in a specific community/area
- Current market prices in a certain city/district
- Direct-from-landlord listings (sublet or direct rent)
- Wanting verified/recent rental data (2024+)

### 2.2 Search Execution Process

**When user requests actual listings or real price data, execute the following:**

#### Step 1: Identify Search Scope

Gather from the user:
- City (required)
- District/Area (required)
- Unit type (optional)
- Budget range (optional)
- Special requirements (e.g., post-2015, direct-from-landlord)

#### Step 2: Execute Multi-Platform Search

**Primary: Xiaohongshu (RED) Search**
- Use `wechat-article-search` skill to search for posts containing:
  - Keywords: `[城市] [区域] 直租`, `[城市] [区域] 转租`, `[小区名] 租房`
  - Filter: Posts from 2024 onwards (request newer posts first)
- Also search: `[城市] [区域] 房东直租 2024`

**Secondary: Web Search for Cross-Validation**
- Use `web_fetch` or search to find:
  - Recent rental listings on Xiaohongshu/Douyin via public posts
  - Cross-reference prices found on multiple platforms

**Supplementary: Douyin (if accessible)**
- Search for rental videos/posts in the target area
- Look for recent posts (2024+) from landlords or tenants

#### Step 3: Collect and Verify Listing Data

For each listing found, record:
- Location (city, district, community name)
- Unit type (1BR, 2BR, etc.)
- Rent price (monthly)
- Year built (if mentioned)
- Direct-from-landlord or sublet
- Source platform + post date
- Contact method (if available)

#### Step 4: Filter Quality Listings

**Include only:**
- Listings from 2024 onwards
- Direct-from-landlord or verified sublet posts
- Complete price information
- Specific location/community named

**Exclude:**
- Listings without prices
-中介 (agent) listings (unless user specifically wants)
- Duplicate posts from same person
- Listings with prices that seem unrealistic

#### Step 5: Generate Report

**Output Template:**

```
## Live Listing Search Results

### Search Parameters
- City: XXX
- Area: XXX
- Unit Type: XXX
- Budget: XXX
- Date Filter: 2024+

### Listings Found: X listings

#### Listing 1: [Community Name / Location]
- Unit Type: XBR
- Rent: XXX/month
- Year Built: XXXX (if mentioned)
- Type: Direct-from-landlord / Sublet
- Source: Xiaohongshu/Douyin (post date)
- Contact: [available or "DM to inquire"]

[Repeat for each listing]

### Price Summary
- Lowest Found: XXX/month
- Highest Found: XXX/month
- Median: XXX/month

### Market Assessment
[Based on actual listings found, how does this compare to the reference data in pricing.md?]

### Recommended Next Steps
1. [Contact the most promising listings]
2. [Verify landlord identity before paying any deposit]
3. [Request video tour if remote]
```

### 2.3 Cross-Platform Price Verification

When actual listings are found, compare against `references/pricing.md`:

| Comparison | Assessment |
|------------|------------|
| Actual price ≤ Reference Low | ✅ Very affordable - verify why (may be sublet, older, shared) |
| Actual price within Reference Range | ✅ Normal - price is consistent with market |
| Actual price 15-30% above Reference High | ⚠️ Premium - check what's included (furnished, metro) |
| Actual price > 30% above Reference High | ❌ Suspicious - verify listing authenticity |

**Important:** If real listings found are significantly different from reference data, trust the live search results and update the report accordingly.

### 2.4 Search Result Currency Notes

- Always prioritize listings from **2024 onwards**
- If no recent listings found, note: "No verified 2024+ listings found in this area. The price estimate below is based on historical data."
- For fast-moving markets, note that prices may have changed since posting

### 2.5 User Interaction Guidance

When user requests search, follow this flow:

```
Step 1: Ask for city + area
Step 2: Ask for optional filters (unit type, budget, year requirement)
Step 3: Execute search across platforms
Step 4: Present findings + price validation
Step 5: Advise on next steps (contact, verify, view)
```

If user asks "Is this rent reasonable?" after seeing actual listings:
→ Use the actual listings found as the baseline, not just the reference data
→ Cross-reference with pricing.md reference data
→ Provide a validated recommendation

---

## Function 3: Price Estimation

### 3.1 Trigger Conditions

User asks:
- Whether rent in a certain area/community is reasonable
- What kind of property can be rented at a certain price
- Reasonable rent range for certain conditions

### 3.2 Estimation Dimensions

| Factor | Impact on Rent |
|--------|----------------|
| Area/District | Core areas have significant premium |
| Metro Distance | Direct metro access > 5 min walk > 10 min walk > further |
| Unit Type/Size | Studios/1BR/2BR/3BR/4BR/Loft tiered pricing |
| Community Quality | Post-2015 new properties > old communities |
| Renovation/Furnishing | Fully furnished > basic > bare shell |
| Floor/Orientation | South-facing > north-facing, mid-to-high floors have premium |

### 3.3 Estimation Principles

- Provide reasonable ranges, not exact figures
- Explain the estimation basis
- Recommend comparing multiple sources before deciding

### 3.4 Output Template

```
## Rent Estimation

### Query Details
- Area/Community: XXX
- Unit Type: XBR
- Size: approx. XX sqm
- Renovation/Furnishing: ...

### Reasonable Rent Range
- Low: XXX/month
- Mid: XXX/month
- High: XXX/month

### Estimation Basis
- Similar listings nearby: XXX/month
- Community quality premium: approx. XX%
- Unit type price differential: ...

### Assessment
[Whether the price is reasonable, expensive, or cheap]
```

### 3.5 Area Price Validation (User Rent Verification)

When needing to understand the user's current rent situation, follow this process:

**Step 1: Ask for Current Rent**

```
What is your current rent? (Area + Community name + Monthly rent)
```

**Step 2: Assess Validity**

After receiving the user's rent information, compare it against the reasonable range for that area:

| Deviation Range | Assessment | Action |
|-----------------|------------|--------|
| Deviation ≤ ±15% | Normal | Record the rent, proceed to next step |
| Deviation 15%-30% | Abnormal | Mark as abnormal, do NOT record, confirm information with user |
| Deviation > 30% | Severely Abnormal | Clearly inform user data is not credible, do NOT record, remind user to verify |

**Step 3: Record After Validation Passes**

Rent data that passes validation can be used for:
- Updating user profile
- Assisting with more suitable area recommendations
- Future price tracking alerts

```
## Rent Record

- User's Area: XXX
- Current Rent: XXX/month
- Validation Status: ✅ Passed
- Recorded At: [Current Time]
```

**Step 4: When Validation Fails**

```
⚠️ The rent information you provided deviates significantly from market rates in that area (approx. XX% deviation).
Please confirm whether the following information is accurate:
- Is the area/community name correct?
- Is the rent the total including utilities?
- Is this your actual rent expenditure?

If there is an error, please re-provide; if confirmed correct, we will skip this recording.
```

---

## Function 4: Contract Review

### Trigger Conditions

User:
- Has a contract and wants help reviewing it
- Asks what to watch out for when signing
- Worried about contract traps
- Dealing with subletting situations

### Review Checklist

#### Must-Verify Items

- [ ] Property certificate and landlord ID match
- [ ] Rent, deposit, and payment method are clearly stated
- [ ] Lease start and end dates are explicit
- [ ] Deposit refund conditions are clearly stated
- [ ] Early termination/subletting terms are reasonable
- [ ] Penalty clauses are not excessive
- [ ] Property inventory list is complete
- [ ] Who bears property fees, utilities, gas

#### High-Risk Clauses (Red Alerts)

1. **"Pay 1 year get 1 year free" / Far below market rent** → Possible scam
2. **No formal contract / Receipt only** → Extremely high risk
3. **No clear deposit refund terms** → Likely to be withheld
4. **Unrestricted subletting allowed** → Possible sublandlord situation
5. **Penalty equals full rent** → Clearly unfair

### Contract Review Output Template

```
## Contract Review Report

### Basic Information Verification
- [ ] Landlord Identity Verified: ✅/❌
- [ ] Property Certificate Verified: ✅/❌
- [ ] Lease Term Verified: ✅/❌

### Key Clause Review

| Clause | Content | Risk Assessment |
|--------|---------|-----------------|
| Rent | XXX/month | ✅ Normal / ⚠️ High / ❌ Abnormal |
| Deposit | XXX | ✅ Normal / ⚠️ High / ❌ Abnormal |
| Payment Method | Monthly/Quarterly/Annual | ⚠️ Note |
| Penalty | XXX | ⚠️ High / ❌ Excessive |

### Risk Alerts
[List identified potential risks]

### Recommended Actions
[Whether to sign / negotiation points / reasons to decline]
```

---

## Tools & Data Sources

### Available Information Retrieval Tools

- **WeChat Article Search**: Use `wechat-article-search` skill to search public account rental articles
- **Web Search**: Use `web_fetch` or exec + curl to scrape public pages
- **Image Recognition**: When user uploads listing screenshots, use `image` tool to analyze

### Reference Data Storage

Local reference data locations:
- `references/shanghai.md` - Shanghai area characteristics and metro info
- `references/beijing.md` - Beijing area characteristics and metro info
- `references/shenzhen.md` - Shenzhen area characteristics and metro info
- `references/guangzhou.md` - Guangzhou area characteristics and metro info
- `references/pricing.md` - Price estimation methodology (live search approach)
- `references/checklist.md` - Contract review checklist detailed version

### How to Use City References

When user provides a city:
1. Load the corresponding city file (e.g., `shanghai.md` for Shanghai)
2. Use area characteristics for context
3. **Always search for live pricing data** — do NOT use static numbers from city files
4. Cross-reference with `pricing.md` methodology for estimation

---

## Notes

1. **Live search is the priority**: When user asks for actual listings or prices, always execute real searches first before falling back to reference data
2. **Prioritize 2024+ data**: Only include listings from 2024 onwards in reports
3. **Cross-validate prices**: Always compare live search results against `references/pricing.md` and flag significant deviations
4. **No legal advice provided**: Contract issues recommend consulting a lawyer
5. **Proactively identify risks**: Clearly warn users when obvious traps are found (e.g., suspiciously low prices, no contract offered)