| name | sg-property-client-report |
| description | Use when a buyer, renter, investor, or property agent asks for Singapore private residential advice, from a broad budget/location/layout brief to a specific condo/project deep dive. Produces an objective client-ready report using sg-housing URA sale/rental evidence plus external project metadata, listings, layout checks, liquidity, peer comparison, and recommendation. |
Singapore Property Client Report
Build a client-ready Singapore private residential report. The user may start with a broad buying brief, a specific project, a layout question, or a project comparison. Treat all of these as one workflow: clarify the client objective, collect evidence, compare alternatives, then give a fair recommendation.
Use this skill for questions like:
- "D18, 1.6-2.2m, 900-1200 sqft, what condos should I consider?"
- "Do a deep dive on D'LEEDON 3-bedders."
- "Is Treasure at Tampines a good buy for a 3-bedder around 1.8m?"
- "Compare The Tapestry and The Alps for self-stay and resale liquidity."
- "For a client with 2.2m budget, which projects should I show and why?"
Core Principle
Do not just return data. Build a defensible recommendation.
Start from transaction evidence, then enrich it with project facts, layout reality, active asking context, location fit, and risks. sg-housing MCP is the evidence base for URA sale/rental transactions; it does not provide bedroom count, unit number, stack, facing, renovation, exact floor plan, exact active listings, or complete unit-mix counts.
Every external project metadata or listing claim must include source name + URL, or be marked needs verification.
Report Modes
Choose the mode from the user brief:
- Buyer brief / shortlist: user gives budget, district/location, size, layout, tenure, MRT/school, self-stay, or investment criteria. Produce a market map, then shortlist 3-5 projects for deeper follow-up.
- Project deep dive: user names a project and asks whether to buy, view, bid, or avoid. Produce a full project report.
- Project comparison: user names 2+ projects. Use the same sections for each, then compare fit and trade-offs.
- Investment/rental check: user cares about rent or yield. Add rental medians/contracts and rough gross yield, with cost caveats.
If the brief starts broad, do not stop at a shallow shortlist. Pick the strongest 2-3 candidates and give enough depth for a client conversation.
Clarifying Questions
Ask at most 1-3 focused questions only if the brief is too broad to bound the search. Do not require every dimension before starting.
Proceed without asking when the user provides enough to form a bounded report, usually:
- budget plus location,
- budget plus size/layout,
- project plus layout/budget/purpose,
- project comparison plus use case.
Useful questions:
- Max budget or comfortable range.
- Target location, MRT, school, or district.
- Size/layout target, for example 3-bedder or 900-1200 sqft.
- Self-stay vs investment.
- Priority: liquidity, entry price, rental yield, schools, commute, long-term hold, or upgrade exit.
State assumptions when proceeding with incomplete input.
Research Workflow
1. Define The Client Lens
Normalize:
- Buyer/renter/investor objective.
- Budget, treating a single number as max budget unless wording says otherwise.
- Location: district, planning area, MRT, school, street, or region.
- Size: convert sqft to sqm using
sqm = sqft / 10.7639 and keep both units in the report.
- Layout: treat bedroom count as desired layout, not a URA field.
- Tenure, age/TOP, property type, rental/yield needs, must-haves, exclusions.
Default windows:
- 12 months for liquidity and current price action.
- 24 months for shortlist screening.
- 24-60 months for longer trend context when sample size is thin.
2. Build The Transaction Base
Use find_private_residential_sale_comparables.
For broad buyer briefs:
- Start with
output_mode: "summary".
- Use filters for district/location, min/max price, min/max area sqm, property type, tenure proxy if available, sale type, and from/to.
- Use
summary.project_summaries to identify candidate projects.
- If sample size is zero, relax one non-critical constraint and explain the retry.
- Fetch rows only for shortlisted projects, capped to a small evidence sample.
For project reports:
- Query the target project first with recent windows.
- Fetch peer projects with similar district/location, age, tenure, unit profile, market segment, or buyer audience.
- Fetch row samples only after summary statistics.
For investment:
- Use
get_private_residential_rental_medians or find_private_residential_rental_contracts.
- Estimate gross yield only when both sale and rent evidence are sufficient.
3. Verify Project And Layout Reality
Use external sources for:
- total units, TOP/completion year, tenure, developer,
- unit mix, floor plans, typical sizes, bedroom count,
- site plan, stack/facing/noise/road exposure where available,
- facilities, maintenance/lifestyle fit,
- MRT, schools, malls, parks, commute,
- active listings and asking prices.
When layout is specified but bedroom count is unavailable in URA:
- Prefer externally verified floor plans and unit mix.
- If using
area_sqm as a proxy, label it exactly: area-band proxy, not verified bedroom count.
- Do not claim exact bedroom transaction count from URA alone.
4. Analyze Liquidity
For layout-specific questions:
layout_turnover_rate = last_12_month_layout_transactions / verified_or_defensible_layout_unit_count
If the layout denominator is unavailable:
project_turnover_rate = last_12_month_project_transactions / total_project_units
Interpretation:
- Below 1% annual turnover: very thin liquidity; transaction evidence is weak.
- 1-3%: low to moderate liquidity.
- 3-6%: healthy resale liquidity.
- Above 6%: active market; check if driven by recent TOP, investor churn, or distressed exits.
Do not present turnover as precise if the denominator is estimated.
5. Price, Peer, And Listing Checks
For price trend:
- Compare latest 12 months vs prior 12 months median price and median PSF.
- Show transaction count for each period.
- If sample size is small, say trend is indicative only.
For peers:
- Pick 3-5 comparable projects.
- Explain why each peer is comparable and where it is imperfect.
- Compare median PSF, median quantum, transaction count, liquidity, tenure/age, and buyer profile.
- Avoid cherry-picking peers to force a conclusion.
For active listings:
- Capture source name, URL, asking price, size, bedrooms, and notable claims.
- Compare asking PSF to recent transacted PSF.
- Treat listings as asking context, not transaction evidence.
- Flag stale, duplicated, or unverifiable listings.
Report Structure
Use this structure for full answers. For short answers, preserve the same logic but compress sections.
-
Executive answer
- Shortlist / watch / pass / proceed to viewing / negotiate only below X.
- Confidence level, transaction window, sample size.
-
Client brief and assumptions
- Budget, location, layout, use case, must-haves.
- Missing inputs and assumptions.
-
Market map or project snapshot
- For broad briefs: candidate pool and why projects were included/excluded.
- For project reports: project, street, district, tenure, TOP, total units, market segment, developer.
- Source every external claim or mark
needs verification.
-
Location and living fit
- MRT/commute, schools, amenities, roads/noise, daily usability.
- Separate verified facts from judgement.
-
Unit mix, layout, and stack/site considerations
- Layout supply, typical sizes, floor-plan fit, site/stack/facing risks if known.
- Mark area-band proxies clearly.
-
Liquidity and turnover
- Layout turnover if defensible; otherwise project turnover proxy.
- Explain what liquidity means for exit risk and price discovery.
-
Price evidence
- Latest range, median price, median PSF.
- 12-month vs prior 12-month movement.
- Evidence rows only if useful, capped at 5-10.
-
Peer comparison
- 3-5 peers with comparability notes.
- Show price, liquidity, tenure/age, and buyer-fit trade-offs.
-
Active listings and negotiation frame
- Current asking evidence if available.
- Asking vs transacted gap.
- Fair range and bid discipline.
-
Recommendation
- What supports buying.
- What argues for caution.
- Who this is suitable for and who should avoid it.
- Due diligence before viewing or offering.
- Sources and caveats
- URA transaction data via sg-housing.
- External source names + URLs.
- URA rows do not include bedroom count, unit number, stack, facing, renovation, or exact layout.
- Decision support, not valuation advice.
Output Rules
- Be objective and fair. Do not write like a sales brochure.
- Do not dump raw JSON.
- Keep tables compact and readable.
- Do not include every matching row or project when the list is long.
- Use SGD and PSF consistently.
- For sqft criteria, show sqm conversion.
- Do not rank solely by cheapest PSF.
- If evidence is thin, say so plainly.
- If external sources conflict, show the conflict instead of choosing silently.
References
Load these only when needed:
references/report-template.md for the full report skeleton.
references/source-map.md for source categories and URL capture checklist.
references/turnover-methodology.md for liquidity calculations.
references/research-patterns.md for Stacked-style depth cues and report dimensions.
Minimum Quality Bar
A good report must include:
- Transaction window and sample size.
- Clear recommendation, not just data.
- Layout/bedroom caveat when layout-specific.
- Liquidity or turnover analysis.
- Price trend and peer comparison.
- Active listing check or explicit statement that it was not completed.
- Source name + URL for external metadata/listing claims, or
needs verification.
- Watch-outs and due diligence.
- Not valuation advice.