SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillscar-advisor

car-advisor

Data & analytics

View repository

Updated 10-15 21:08

About this skill

Car Advisor Skill

An automotive Q&A and comparison system based on real-time data. Core principles:Don't guess with training data, speak with real-time retrieval。

Workflow Overview

用户提问
   ↓
[Step 1] 解析意图 → 确定车型、对比维度、问题类型
   ↓
[Step 2] 并行数据采集 → 参数 + 价格 + 真实评论
   ↓
[Step 3] 数据整理 → 生成结构化表格或分析报告
   ↓
[Step 4] 输出结论 → 附数据来源、注明采集时间

Step 1: Intent Parsing

After receiving the user's question, first identify internally:

Dimension Extracted Content
Vehicle Model List Which specific vehicle models are involved (brand + model line + model year)
Question Type Specification inquiry / side-by-side comparison / reputation reviews / car purchase advice / price analysis
Core Concerns The dimensions the user cares about most (range/intelligent driving/interior/price/space…)
Data Freshness Requirements Whether the latest prices/latest configurations are needed (such as refreshed versions, new models)

If the vehicle model information is ambiguous, you can first use web_search to confirm the specific version currently on sale (for example, the official name of "Model Y 2025 model").

Step 2: Data Collection Strategy

2.1 Data Source Priority

For the detailed data source list, see references/data-sources.md。

Priority 1 — Official pages (most authoritative)

  • Tesla:https://www.tesla.cn/modely、https://www.tesla.cn/model3
  • Xiaomi:https://www.xiaomiev.com
  • AITO:https://www.aito.com
  • Li Auto:https://www.lixiang.com
  • BYD/domestic brand official websites

Priority 2 — Vertical media specification databases (most comprehensive)

  • Dongchedi specification pages:https://www.dongchedi.com/auto/params-carIds-x-{car_id}
  • Autohome specification comparison:https://www.autohome.com.cn/config/
  • Use web_search to search for "model name specifications and configuration Dongchedi" to quickly locate the page

Priority 3 — Real comments/reputations

  • Dongchedi reputation: search {车型名} 口碑 懂车帝
  • Autohome owner reviews: search {车型名} 车主评价 汽车之家
  • Zhihu car usage reports: search {车型名} 用车体验 知乎
  • Media reviews: search {车型名} 深度评测 2024

Priority 4 — Price/sales data

  • Current selling price: search {车型名} 最新价格 2025(prices change frequently and must be real-time)
  • Sales data: search {车型名} 销量 最新

2.2 Parallel Retrieval Strategy

For comparisons of multiple models, initiate multiple searches in parallel; do not wait serially. Typical search combinations:

同时搜索：
- "{车型A} 配置参数 {年款}"
- "{车型B} 配置参数 {年款}"
- "{车型A} vs {车型B} 对比 评测"
- "{车型A} 车主评价 缺点"
- "{车型B} 车主评价 缺点"

Step 3: Data Organization Standards

3.1 Specification Comparison Table Template

When generating a comparison table, adjust the dimensions dynamically according to the user's concerns, but it is recommended to include the following core dimensions:

Basic Information Layer

Dimension Description
Model/Version Brand + model line + specific version + model year
Official starting price Indicate whether subsidies are included, and note the collection date
Body dimensions Length × width × height, wheelbase
Positioning Class (A/B/C class), body style

Power/Range Layer (key for new energy vehicles)

Dimension Description
Battery capacity kWh
CLTC range km
Maximum power kW/hp
0-100 km/h acceleration seconds
Charging power kW (peak fast charging)
Energy consumption kWh/100km

Interior/Comfort layer

Dimension Description
Seat configuration Heating/ventilation/massage/power adjustment, specify front and rear rows
Central control screen Size + system
Rear screen Yes/no, size
Infotainment system Brand/version, phone connectivity capability
Ambient lighting Yes/no, number of zones
Audio Brand + number of speakers
Air suspension Yes/no
Special features Fragrance/refrigerator/V2L/camping mode

Intelligent driving layer

Dimension Description
Intelligent driving solution Brand/version
LiDAR Yes/no
Highway NOA Whether supported and fee situation
City NOA Whether supported and number of cities covered
Intelligent driving hardware Computing power TOPS

Safety/Space

Dimension Description
Number of airbags units
Trunk volume L (including front trunk)
Flat-folding Whether supported, length

3.2 Review integration specifications

After the comparison table, attach a summary of real reviews, in the following format:

## 车主真实反馈

### [车型A]
**好评高频词**：[提炼3-5个]
**投诉集中点**：[提炼3-5个]
**代表性车主评论**：
- "..." (来源：懂车帝/汽车之家，评分X星)
- "..."

### [车型B]
...

> 数据来源：懂车帝口碑、汽车之家车主点评
> 采集时间：{当前日期}
> 注：评论为用户主观表达，仅供参考

Step 4: Output specifications

4.1 Comparison conclusion structure

## 一句话总结
[直接给出核心结论，不废话]

## 参数对比表
[完整表格]

## 各维度胜出分析
- 续航/能效：XX胜出，原因...
- 智驾：XX胜出，原因...
- 内饰/舒适：XX胜出，原因...
- 性价比：XX胜出，原因...

## 车主真实反馈
[评论摘要]

## 购买建议
如果你更在意[X]，选[车型A]
如果你更在意[Y]，选[车型B]

> 数据采集时间：{日期} | 价格以官网最新为准

4.2 Single model query structure

## [车型名] 当前配置一览

[关键参数，重点回答用户问题]

## 各版本差异
[如有多版本，列出关键差异]

## 真实车主评价
[评论摘要]

## 综合评价
[一段客观评述]

Special scenario handling

Car purchase budget recommendations

When the user provides a budget range:

  1. First search for currently popular models in that budget range (search "200,000-300,000 RMB new energy SUV 2025 recommendations")
  2. Filter out 3-5 candidates and make a brief comparison
  3. Give ranking suggestions according to the preferences the user explicitly mentions (family/sporty/long-distance/urban)

Price change query

Price is highly dynamic information (Tesla especially adjusts prices frequently):

  • Must search in real time; do not use prices from training data
  • When searching, add time words such as "latest" and "2025" to ensure information freshness
  • If search results show disputed prices, note "subject to the official website in real time"

When data cannot be obtained

If a specific parameter cannot be confirmed through search:

  • Clearly state "this parameter could not be confirmed from real-time data"
  • Provide reference source links so users can verify on their own
  • Do not fill in with training data (may be outdated)

Data quality principles

  1. Timeliness first: Always add a year word when searching ("2025" or "latest") to ensure the information is not an outdated version
  2. Cross-validation: For important parameters (price, range), try to confirm from 2 sources
  3. Source transparency: Each key data point indicates its source (official website/Dongchedi/Autohome)
  4. Separation of subjective and objective: The parameter table uses objective data, while the comment section uses subjective feedback; the two are not mixed
  5. Honesty about uncertainty: When data is disputed, explain it, and do not force a definitive answer

Reference Resources

  • For a detailed list of data sources, see references/data-sources.md
  • For the comparison dimension priority matrix, see references/compare-dimensions.md