IDX CMA Report
Use this skill to turn subject-property data and IDX comparables into a defensible CMA package with:
- - Structured valuation calculations
- A written report for agent/client review
- An interactive handoff prompt for Google Gemini Canvas / Google AI Studio
Workflow
1. Gather Data Through IDX MCP/CLI
Use the IDX MCP/CLI skill already available in the environment to pull:
- - Subject property details
- Candidate comparable listings (closed/pending/active based on user preference)
Ask the user which comps to include when the choice is ambiguous. Keep 3 to 8 comps unless the user requests otherwise.
Normalize data to JSON using the schema in references/cma-input-schema.md.
2. Build CMA Outputs
Run:
CODEBLOCK0
The script produces:
- -
cma-output/cma_report.md (summary report) - INLINECODE2 (calculation payload)
- INLINECODE3 (local interactive view)
- INLINECODE4 (prompt for Google tools)
3. Review and Explain Adjustments
Before final delivery:
- - Show the comp set used
- Show estimated range and central estimate
- Explain assumptions and major adjustments in plain language
- Flag missing/low-quality fields that weaken confidence
Use references/valuation-guidelines.md for adjustment defaults and confidence guidance.
4. Publish Interactive Version in Gemini
Use
cma-output/gemini_canvas_prompt.md as the base prompt. Then:
- 1. Open Google AI Studio or Gemini Canvas.
- Paste the generated prompt and provide
cma_data.json. - Ask for an interactive CMA web app with:
- Comp table with sorting/filtering
- Map-ready data fields (if lat/lng present)
- Value-range visualization
- Notes panel explaining adjustments
- 4. Request hosted/shareable output if available in the chosen Google tool.
See references/gemini-canvas-publish.md for a copy-ready checklist.
Safety Rules
- - Treat outputs as broker/agent CMA support, not a licensed appraisal.
- Surface data gaps, outliers, or stale comps before presenting a valuation.
- Never invent listing attributes; mark missing values as unknown.
- Keep a clear boundary between factual listing data and model assumptions.
References
- - INLINECODE9
- INLINECODE10
- INLINECODE11
IDX CMA 报告
使用此技能将主题房产数据和IDX可比房源转化为可辩护的CMA包，包含：
- - 结构化估值计算
- 供经纪人/客户审阅的书面报告
- 面向Google Gemini Canvas / Google AI Studio的交互式交接提示
工作流程
1. 通过IDX MCP/CLI收集数据
使用环境中已有的IDX MCP/CLI技能获取：
- - 主题房产详情
- 候选可比房源列表（根据用户偏好选择已成交/待定/在售状态）
当选择存在歧义时，询问用户应包含哪些可比房源。除非用户另有要求，保留3至8个可比房源。
使用references/cma-input-schema.md中的架构将数据标准化为JSON格式。
2. 构建CMA输出
运行：
bash
python3 scripts/build_cma.py \
--subject subject.json \
--comps comps.json \
--output-dir cma-output
该脚本生成：
- - cma-output/cmareport.md（摘要报告）
- cma-output/cmadata.json（计算数据负载）
- cma-output/interactivelocal.html（本地交互视图）
- cma-output/geminicanvas_prompt.md（面向Google工具的提示）
3. 审阅并解释调整项
在最终交付前：
- - 展示所使用的可比房源集合
- 展示估值区间和中心估值
- 用通俗语言解释假设条件和主要调整项
- 标记缺失/低质量字段以提示置信度不足
使用references/valuation-guidelines.md获取调整默认值和置信度指导。
4. 在Gemini中发布交互版本
使用cma-output/gemini
canvasprompt.md作为基础提示。然后：
- 1. 打开Google AI Studio或Gemini Canvas。
- 粘贴生成的提示并提供cmadata.json。
- 请求创建交互式CMA网页应用，包含：
- 可排序/筛选的可比房源表格
- 地图就绪数据字段（如存在经纬度）
- 价值区间可视化
- 解释调整项的备注面板
- 4. 如所选Google工具支持，请求生成可托管/可分享的输出。
参见references/gemini-canvas-publish.md获取可直接使用的检查清单。
安全规则
- - 将输出视为经纪人/代理商的CMA支持工具，而非持牌评估。
- 在呈现估值前，揭示数据缺口、异常值或过时可比房源。
- 绝不虚构房源属性；将缺失值标记为未知。
- 在事实性房源数据与模型假设之间保持清晰界限。
参考资料
- - references/cma-input-schema.md
- references/valuation-guidelines.md
- references/gemini-canvas-publish.md
