NodeWeave

Researched and written 2026-09-29 14:19 SGT · 19 sources listed at the end · draft 1 · figures verified against the named sources on the date shown; where a source has since moved, the element carries a marker.

You are reading a proof page, not a door. Nothing here is for sale and no price is published. This page exists because you are a licensed real estate salesperson deciding whether the rest of this site — the fact ledger, the citations, the conflict markers, the twelve-question frame repeated on every page — is a genuine piece of infrastructure or a very consistent-sounding piece of marketing copy. The honest answer is that it is infrastructure, it is owned entirely by one operator, and this page is the one place on the site that turns the machine around to look at itself instead of at a property.

The mechanics: what NodeWeave actually is

NodeWeave is not a chatbot bolted onto a real estate website. It is a fact ledger, a page writer, and a verifier, in that order, and none of the three trusts the other two. Every figure that appears anywhere on this site — a stamp duty band, a loan-to-value ceiling, a probate deadline — exists first as a row in a database table: a dotted id, a value, a unit, the government body or publication it came from, and the date it was read. A page is written only after that row exists or is created alongside it, and a page element cites the row's id rather than retyping the number. That single design choice is the whole difference between this site and a property blog: a property blog states a figure once and it goes stale silently; here, a figure changing anywhere is a database event a page can be checked against.

The shelf behind this page holds 1,792 facts as of the day this page was written, spanning everything from Additional Buyer's Stamp Duty bands to executor commission caps to industrial loan-to-value practice. Roughly one in six of those facts was read directly off the regulator's own page by a scheduled ingestion process — marked [LINE] internally — which outranks a figure a page writer extracted secondhand from a summary of that same regulator. The rest were fetched, dated, and sourced the same way this paragraph was: government first, property press and market guides only where government would not render a figure on the day it was read.

When two sources disagree on a figure, the system does not quietly pick one. A conflicting value is recorded as a conflict row, the fact is marked contested, and any page citing it renders with a visible marker naming both figures until a human rules on it — a ladder resolves the easy cases automatically (a regulator's own direct-read line beats a page writer's extraction of the same regulator; a body's own later instrument beats its earlier one), and only a genuine disagreement between two equally authoritative sources reaches the operator's desk. You can see this in the ledger today: the non-residential Buyer's Stamp Duty bands are marked as unreconciled between two sources rather than silently resolved in whichever direction reads better. That is not a bug this page is hiding — it is the mechanism working as built.

A page like this one is produced by a fixed procedure, not by a person improvising: read the page's own brief from the site's sitemap row, search the shelf for what is already held rather than re-fetching it, fetch only genuine gaps from government sources first, write the JSON, run it through a writer that checks every fact against the ledger before it touches a live page, then read the page back through the same public rendering path a visitor uses and refuse to call the job done unless the heading, the element count, and every one of twelve required spokes actually survived the round trip. This page cleared that same floor: at least 22 rendered elements, at least 1,200 words, and coverage across eligibility, tax, borrowing, conveyancing, estate, the twelve-question frame, the B2B section, and the realtor economics section below — checked mechanically against the page as it actually rendered, not against what a model claimed it wrote.

Eligibility: who this machine is actually built for

NodeWeave does not lower the bar the Council for Estate Agencies sets, and it is not built to let anyone skip it. Since 1 January 2011 estate agency work in Singapore — residential, commercial or industrial alike — has required a CEA licence for the agency and CEA registration for the individual salesperson doing the work; doing it without registration is an offence carrying a fine of up to S$75,000, up to three years' imprisonment, or both. Registering at all requires at least four GCE O-Level passes or the equivalent, a pass in both papers of the Real Estate Salesperson examination from a CEA-approved provider, professional indemnity insurance running at least a year, current CPF MediSave contributions, and a minimum age of 21. A salesperson registers through exactly one licensed agency at a time.

The reader this page is written for is a solo practitioner: someone already CEA-registered, running their own client base without a large agency's marketing department behind them, who is looking at this site's output and asking whether the same machine could sit behind their own name. NodeWeave writes pages and holds a fact ledger; it does not register anyone with CEA, does not renew anyone's licence — the current 2026 renewal cycle runs 1 October to 30 November, with a late window to 31 December that carries a real risk of lapsing into the new year if missed — and it does not make an unregistered person's output lawful to sell. The machine can write a correct, cited, dated page. Whether the person publishing it under their own name is allowed to charge a fee for the transaction the page describes is a CEA question NodeWeave has no authority over, and it never pretends otherwise.

Citizenship status: proof the machine holds the distinction, not just states it

The clearest test of whether a system like this actually works, rather than sounding like it does, is whether it keeps four different citizenship profiles straight on the single tax line that punishes confusing them. Additional Buyer's Stamp Duty is a ladder set by who a buyer is, not by what they can afford: a Singapore Citizen pays nothing extra on a first home, 20% on a second, 30% on a third and beyond; a Singapore Permanent Resident pays 5% on a first home, 30% on a second, 35% on a third; a foreigner pays 60% on any residential property; an entity — a company, a trustee, a partnership acting as one — pays 65% regardless of which property it is. Get any one of those four wrong on a client's behalf and the error is measured in tens or hundreds of thousands of dollars, not in an awkward correction.

Five nationalities sit outside that ladder by treaty rather than by citizenship: nationals of the United States, Switzerland, Liechtenstein, Norway and Iceland are taxed at the Singapore Citizen rate under Singapore's free trade agreements with those states, and this runs through even where a US citizen also holds Singapore Permanent Residency — the FTA rate applies, not the PR rate. The ledger holds this as a stamp-duty rule only: an FTA national buying a landed house or vacant residential land still needs the same Land Dealings Approval Unit approval any other foreigner needs, because that is a different statute answering a different question, and the machine does not collapse the two the way a careless summary would.

Taxes: what the ledger holds, dated, and what it refuses to compute for you

Buyer's Stamp Duty on residential property runs a fixed progressive schedule: 1% on the first S$180,000, 2% on the next S$180,000, 3% on the next S$640,000, 4% on the next S$500,000, 5% on the next S$1,500,000, and 6% on the remainder above S$3,000,000, rounded down to the nearest dollar. Seller's Stamp Duty on residential property bought on or after 4 July 2025 runs a four-year holding period at 16% in year one, 12% in year two, 8% in year three, 4% in year four, nil after — a longer window and higher rates than the 12/8/4/0 over three years that applied before that date, and several market guides still quote the old schedule, which is exactly the kind of staleness the ledger's own staleness check exists to catch. Property tax on an owner-occupied home runs from 0% on the first S$12,000 of Annual Value up to 32% above S$140,000; a non-owner-occupied residential property runs a flatter, steeper schedule from 12% up to 36%.

What NodeWeave will not do is compute a specific client's stamp duty, property tax or rental income tax exposure and present it as an answer. IRAS's own Buyer's Stamp Duty, Additional Buyer's Stamp Duty and property tax pages frequently return a navigation shell rather than the figure itself on a direct fetch, which the ledger records as a gap rather than papering over with a guess — a wrong number with a licensed salesperson's name on it is a worse outcome than an acknowledged gap pointing at IRAS's own calculator, because the arithmetic depends on facts about the specific buyer — citizenship, property count, holding structure — that a general page can never see.

Borrowing: the machine tracks the ceiling, it does not set it or lend against it

MAS sets loan-to-value limits that the ledger holds and checks pages against, not the other way round: a first housing loan carries a 75% LTV ceiling (55% if tenure runs past 30 years, 25 for an HDB flat, or the loan runs past age 65), a second outstanding loan drops to 45%, a third and beyond to 35%, and a non-individual borrower — a company — is capped at 15% regardless. Total Debt Servicing Ratio sits at 55% of gross monthly income, assessed at a MAS floor rate of 4% for residential property regardless of the rate actually quoted, specifically so a low headline rate cannot buy a borrower more loan than the floor allows; the Mortgage Servicing Ratio caps HDB and EC purchases at 30% during the minimum occupation period. HDB cut its own loan-to-value limit from 80% to 75% in August 2024 to match the bank limit — a rule change the ledger picked up because the fact's effective date moved, which is precisely the kind of change a page citing the old 80% figure is designed to be flagged as stale against and rewritten for, not left quietly wrong.

NodeWeave is not a bank, a licensed moneylender, or a mortgage broker, and it holds no position in any transaction it writes about. An HDB loan and a bank loan differ on more than the headline LTV: an HDB loan's 25% downpayment can be paid entirely in cash, CPF Ordinary Account, or a mix, with no compulsory cash portion; a bank loan's downpayment requires at least 5% in cash regardless of how much CPF a buyer holds. The ledger tracks both sides of that split as separate, dated facts rather than one blended figure, because collapsing them is exactly the kind of merge that produces a wrong answer for whichever buyer happens to be on the HDB side of it.

Renovation, addition and alteration, and rebuild: coverage as a demonstration of breadth

This site's reno and A&A pages sit outside the taxes-and-financing spine, and they exist in the ledger anyway, which is itself a small proof of how far the machine's coverage runs. Where a landed house already exceeds its permissible building envelope, URA assesses further addition-and-alteration work on its merits rather than granting it automatically, and where an owner retains a non-conforming structure, URA asks for a Professional Engineer's certification that it remains structurally sound before anything proceeds. On the financing side, a standalone bank renovation loan is reported running around 4.2% to 4.9% flat (an effective rate closer to 6% once fees are amortised over a shorter tenure), against a 2-year fixed mortgage rate that can be roughly four times cheaper for the same principal over five years — a gap the ledger holds as a sourced, dated comparison rather than a rule of thumb, with the disagreement between two market guides on the exact headline rate stated plainly rather than resolved by picking whichever number sounds more confident.

Conveyancing: the machine coordinates the sequence, a lawyer still does the legal work

NodeWeave does not draft, review or file a single conveyancing document, and no page it writes claims to. What the ledger does hold is the sequencing information a solo practitioner actually needs in the moment: CPF Board describes the practical order for stamp duty as pay Buyer's and Additional Buyer's Stamp Duty upfront in cash to meet the 14-day deadline, then apply for CPF reimbursement once a CPF charge is lodged on the property, because CPF cannot disburse before its own charge is registered — a sequencing fact, not a legal opinion, and exactly the kind of thing a fact ledger can hold precisely where a page written once and never revisited tends to go vague. HDB's own in-house conveyancing fee scale for a resale buyer runs 13.5 cents per S$100 on the first S$30,000, 10.8 cents per S$100 on the next S$30,000, and 9 cents per S$100 on the remainder, rounded up before GST — a schedule, not an estimate, and one the ledger cites by id rather than restating from memory on every page that needs it.

Estate: the same ledger, applied to a harder category of client

An estate transaction is where a fact ledger earns its keep hardest, because the reader is often grieving, unfamiliar with the process, and dealing with a hard deadline at the same time. The ledger holds the gate most agents skip past: a Grant of Probate is available only where a valid will exists naming the applicant executor; without one, the route is Letters of Administration instead. It holds the deadline that actually bites: an HDB estate flat must be sold within six years of the death. And it holds the figure worth setting next to what a selling agent earns on the same transaction: a court may award an executor a commission of up to 5% of the assets they collected for the work of administering the estate. [⚠ estate.probate.eligibility_2026_06_15: sources differ; awaiting a ruling]

The twelve questions, applied to adopting this machine into your own practice

This is the frame used on every page on this site, and here it turns on the decision underneath this whole page: whether a solo practitioner should want a system like this sitting behind their own name. Two questions sort every reader before the twelve start: are you borrowing to buy, and once you hold it, are you lending it out? Adopting NodeWeave sits outside that 2x2 the same way hiring an agent does — it changes neither what you own nor what you owe. What it changes is what a correct, cited, dated page costs you to produce, and how fast, which is exactly what the twelve answers below work through.

C0. The play: balance sheet or cash flow?

Adopting a system like this is a cash-flow decision, not a balance-sheet one. A solo practitioner does not own anything different for running it. What changes is what moves through their business month to month — the cost of producing and maintaining correct, cited content against the alternative cost of a marketing retainer, a freelance writer, or simply not publishing at that depth at all.

C1. Balance sheet: what you own and what you owe

Nothing about running this kind of system appears on a practitioner's balance sheet. It produces no asset that sits on the books and no liability that recurs there either; a licensing or engineering cost, if one existed, would be an operating expense of running the practice, not a line that changes what the practitioner owns or owes.

C2. Cash flow: what actually moves, and when

This page carries no published price, because none exists yet — the operator has stated plainly that nothing on this site is for sale and nothing here has been signed off. If that changes, the honest cash-flow line would be a recurring cost against an uncertain future benefit: content production or subscription cost paid on a schedule, against commission income that may or may not follow from it, in a market where CEA sets no fixed commission rate and a seller conventionally pays about 2% on a private resale.

C3. P&L: what it does to profit, which is not the same as cash

A recurring content or licensing cost, if one existed, would sit against revenue as an ongoing operating expense — closer to a marketing line than to a one-off transaction cost like an agent's commission. Confusing a recurring cost of running a practice with a one-off cost of closing a single deal is exactly the kind of merge this frame exists to prevent, and it matters more here than on a single-transaction page because the cost, unlike a commission, does not stop the day one deal closes.

C4. Carry: income minus financing cost

The carry question for a solo practitioner is whether the content this machine produces converts into enough additional trust, enquiry volume, or closed transactions to outrun its own ongoing cost. That is a real, measurable carry — positive if the pipeline it generates outpaces what it costs to run, negative if it becomes a retainer paid on faith. Nothing on this page claims a number for that; it has not been measured publicly, and this page will not invent one.

C5. Give and get: the trade, stated plainly

What a practitioner would give is whatever a licence or engineering cost turns out to be, once one is published — this page states none. What a practitioner would get is a fact ledger that flags its own staleness, a page writer that cannot cite a figure it cannot source, and a verification step that refuses to call a page done until it actually renders correctly to a real visitor — the same discipline this instance holds itself to, extended to their own name instead of Stanley Tan's.

C6. Exposed to: what can move against you

A practitioner running this kind of system is exposed to the machine's own correctness — a fact that has moved and not yet been caught, a source that renders a navigation shell instead of a figure and gets logged as a gap rather than guessed — and separately to their own CEA standing, which the machine has no authority over and does not touch. Those are two different risks, and this page has kept them apart on purpose: a wrong figure is an engineering failure to catch and fix; an unregistered person publishing under a licensed-sounding page is a compliance failure the machine cannot absorb on anyone's behalf.

C7. For how long: the horizon you are committing to

There is no term to commit to, because there is no contract offered on this page. If a practitioner reaches out through work-with-me and something is eventually agreed, the horizon would be whatever that agreement states — this page does not presume an answer it has not been given.

C8. Failure cascade: if it goes wrong, in what order

If a recurring content cost outruns what it generates, the cascade runs the same order this frame always uses: negative carry hits cash flow first, then accumulates into the practice's P&L if it persists across enough months. It does not reach negative equity or insolvency through the system alone, because a content or licensing cost, unlike a leveraged property position, carries no collateral and no compounding balance — the practitioner can simply stop paying for it, which is a materially different failure mode from the ones this frame usually describes for a property itself.

C9. Liquidity: how fast, and at what haircut

This is the sharpest contrast with almost every other page on this site: a property position is illiquid by nature and can take months to convert back to cash, often at a haircut. A content or software arrangement, if one existed, would be close to fully liquid — a practitioner can stop using it on notice, with none of the transaction costs, stamp duties, or market-timing risk a property sale carries.

C10. Control: can you move the outcome, or are you a price-taker?

This is the spoke this page exists to make a case for. A solo practitioner buying generic SEO content from a marketing vendor is a price-taker on quality, sourcing and accuracy — the vendor decides what gets published and the practitioner's name is the one exposed if it is wrong. A ledger-backed system gives the practitioner more control over what is actually published under their own licence: every figure traceable to a dated source, every conflict visible rather than hidden, every stale fact flagged for rewrite rather than left to quietly drift. Control here means being able to check the page before a client does, not merely publishing more pages faster.

C11. The exit: how you get out, and what it costs

There is nothing to exit, because nothing has been entered into — this page is a demonstration, not a signed relationship. If a commercial arrangement is ever offered, the honest expectation, stated in advance rather than discovered later, would be that the exit terms — what happens to already-published pages, whether they stay live under the practitioner's name or come down — are the kind of question this page would answer plainly once there is an answer to give, not the kind it would leave implicit.

The B2B spoke: what this page is actually offering a solo practitioner

Said plainly: this page is not a storefront. No price is published, no contract is offered, and no claim is made here that NodeWeave is currently available to license. What this page is, is proof — that a fact ledger, a page writer, and a verification loop can hold 1,792 facts across dozens of property topics, catch their own conflicts, flag their own staleness, and still produce a page that reads like a person wrote it and checked their sources, because that is exactly what happened, on every page including this one. If you are a CEA-registered solo practitioner and this is the kind of infrastructure you would want behind your own name, the honest next step is to say so directly rather than for this page to pretend a commercial offer exists when the operator has not made one.

What this page will not do is dress up a software pitch as financial or investment advice, or let a claim about lead generation or conversion substitute for a verifiable figure — the same Financial Advisers Act boundary that keeps a property agent from doubling as an unlicensed adviser applies with equal force here: a claim about what a tool does for a practitioner's business has to be true and checkable, not just plausible.

The realtor economics: what this changes about what a practitioner earns, and what it doesn't

CEA sets no fixed commission rate and no prescribed guideline on the amount, for any property type — the figure is agreed between agent and client before work starts, in a Prescribed Estate Agency Agreement, and market convention on a private resale has the seller paying about 2%. Nothing about a content system, this one or any other, changes that structure: a practitioner using NodeWeave still earns the same negotiated commission, on the same terms, paid the same way — at completion, through the licensed agency, never split between both sides of a transaction. What a system like this changes is not the commission mechanism; it is the cost and reliability of the content that brings a client to the point of engaging an agent at all.

Where the incentive points, plainly, including against the operator: a system that writes property content faster and cheaper than a person doing it by hand has an obvious incentive to publish more of it, faster, whether or not each page clears the bar this one claims to — which is exactly why the verification step exists as a mechanical floor rather than a promise, and exactly why this page states its own limits (no price, no contract, no lead-generation figure) instead of implying more certainty than the operator actually has. A tool that makes content cheap to produce is not automatically a tool that makes content correct; this page's own claim to correctness rests on the ledger and the verification loop being checkable by anyone, not on being taken on faith.

What this page will not claim

• It will not state a price, a contract term, or a commercial offer that does not exist — this is a proof page, and the operator has signed off nothing.
• It will not claim a lead-generation, conversion or income figure it cannot source — no such figure has been measured publicly, and none is invented here.
• It will not imply that running this system replaces CEA registration, renewal, or the licensing conditions any solo practitioner must meet independently.
• It will not dress up a software or content pitch as financial or investment advice.
• It will not claim a fact is settled when the ledger itself holds it as contested — a visible conflict marker is the honest output, not a bug to hide.
• It will not let a stale figure sit on this page after the rule behind it has changed; the same staleness check that applies to every other page on this site applies to this one.

Sources

• CEA — Information for aspiring real estate salespersons (2026-03-03) · https://www.cea.gov.sg/real-estate-professionals/for-aspiring-real-estate-salespersons/information-for-aspiring-real-estate-salespersons/
• CEA — Renewal of Salesperson Registration (2026-10-01) · https://www.cea.gov.sg/real-estate-professionals/for-real-estate-salespersons/renewal-of-salesperson-registration/
• CEA — Estate Agency Fees (2026-05-13) · https://www.cea.gov.sg/consumers/estate-agency-fees
• IRAS — Additional Buyer's Stamp Duty (ABSD) (2023-04-27) · https://www.iras.gov.sg/taxes/stamp-duty/for-property/buying-or-acquiring-property/additional-buyer's-stamp-duty-(absd)
• IRAS — Buyer's Stamp Duty (BSD) rates (2023-02-15) · https://www.iras.gov.sg/taxes/stamp-duty/for-property/buying-or-acquiring-property/buyer's-stamp-duty-(bsd)
• IRAS — Foreigners Eligible for ABSD Remission under Free Trade Agreements (FTAs) (2026-09-28) · https://www.iras.gov.sg/taxes/stamp-duty/for-property/appeals-refunds-reliefs-and-remissions/common-stamp-duty-remissions-and-reliefs-for-property/foreigners-eligible-for-absd-remission-under-free-trade-agreements-(ftas)
• MAS — Loan Tenure and Loan-to-Value Limits (2024-03-27) · https://www.mas.gov.sg/regulation/explainers/new-housing-loans/loan-tenure-and-loan-to-value-limits
• MAS — Calculating TDSR for Property Loans (2018-07-05) · https://www.mas.gov.sg/regulation/explainers/tdsr-for-property-loans/calculating-tdsr
• MAS — MSR and TDSR for property loans (2021-12-16) · https://www.mas.gov.sg/regulation/explainers/new-housing-loans
• MAS — Seller's Stamp Duty for residential property, four-year schedule (2025-07-04) · https://www.mas.gov.sg
• HDB — Measures to Cool the HDB Resale Market and Provide More Support for First-Time Home Buyers (2024-08-20) · https://www.hdb.gov.sg/cs/infoweb/about-us/news-and-publications/press-releases/19082024-Measures-to-Cool-the-HDB-Resale-Market-and-Provide-More-Support-for-First-Time-Home-Buyers
• URA — Additions and Alterations to Landed Housing (2026-06-07) · https://www.ura.gov.sg/guidelines/development-control/development-control-handbooks/residential/bungalows/aa-landed-housing/
• CPF Board — HDB vs bank loan downpayment (2026-06-12) · https://www.cpf.gov.sg
• CPF Board — Stamp duty reimbursement mechanics (2026-04-22) · https://www.cpf.gov.sg
• Family Justice Courts — Apply for probate (2026-06-15) · https://www.judiciary.gov.sg/family/apply-for-probate
• SingaporeLegalAdvice.com — Executor commission cap (2025-01-14) · https://singaporelegaladvice.com
• TRIBE — Bank renovation loan rates and effective cost comparison (2026-09-16) · https://tribe.com.sg
• NodeWeave — spine/portal_page.py, the page-writer's own source, read directly this run (2026-09-29) · spine/portal_page.py
• NodeWeave — spine/portal_write.py, the writer's own source, read directly this run (2026-09-29) · spine/portal_write.py

Sources

• CEA — Information for aspiring real estate salespersons (2026-03-03) · https://www.cea.gov.sg/real-estate-professionals/for-aspiring-real-estate-salespersons/information-for-aspiring-real-estate-salespersons/
• CEA — Renewal of Salesperson Registration (2026-10-01) · https://www.cea.gov.sg/real-estate-professionals/for-real-estate-salespersons/renewal-of-salesperson-registration/
• CEA — Estate Agency Fees (2026-05-13) · https://www.cea.gov.sg/consumers/estate-agency-fees
• IRAS — Additional Buyer's Stamp Duty (ABSD) (2023-04-27) · https://www.iras.gov.sg/taxes/stamp-duty/for-property/buying-or-acquiring-property/additional-buyer's-stamp-duty-(absd)
• IRAS — Buyer's Stamp Duty (BSD) rates (2023-02-15) · https://www.iras.gov.sg/taxes/stamp-duty/for-property/buying-or-acquiring-property/buyer's-stamp-duty-(bsd)
• IRAS — Foreigners Eligible for ABSD Remission under Free Trade Agreements (FTAs) (2026-09-28) · https://www.iras.gov.sg/taxes/stamp-duty/for-property/appeals-refunds-reliefs-and-remissions/common-stamp-duty-remissions-and-reliefs-for-property/foreigners-eligible-for-absd-remission-under-free-trade-agreements-(ftas)
• MAS — Loan Tenure and Loan-to-Value Limits (2024-03-27) · https://www.mas.gov.sg/regulation/explainers/new-housing-loans/loan-tenure-and-loan-to-value-limits
• MAS — Calculating TDSR for Property Loans (2018-07-05) · https://www.mas.gov.sg/regulation/explainers/tdsr-for-property-loans/calculating-tdsr
• MAS — MSR and TDSR for property loans (2021-12-16) · https://www.mas.gov.sg/regulation/explainers/new-housing-loans
• MAS — Seller's Stamp Duty for residential property, four-year schedule (2025-07-04) · https://www.mas.gov.sg
• HDB — Measures to Cool the HDB Resale Market and Provide More Support for First-Time Home Buyers (2024-08-20) · https://www.hdb.gov.sg/cs/infoweb/about-us/news-and-publications/press-releases/19082024-Measures-to-Cool-the-HDB-Resale-Market-and-Provide-More-Support-for-First-Time-Home-Buyers
• URA — Additions and Alterations to Landed Housing (2026-06-07) · https://www.ura.gov.sg/guidelines/development-control/development-control-handbooks/residential/bungalows/aa-landed-housing/
• CPF Board — HDB vs bank loan downpayment (2026-06-12) · https://www.cpf.gov.sg
• CPF Board — Stamp duty reimbursement mechanics (2026-04-22) · https://www.cpf.gov.sg
• Family Justice Courts — Apply for probate (2026-06-15) · https://www.judiciary.gov.sg/family/apply-for-probate
• SingaporeLegalAdvice.com — Executor commission cap (2025-01-14) · https://singaporelegaladvice.com
• TRIBE — Bank renovation loan rates and effective cost comparison (2026-09-16) · https://tribe.com.sg
• NodeWeave — spine/portal_page.py, the page-writer's own source, read directly this run (2026-09-29) · spine/portal_page.py
• NodeWeave — spine/portal_write.py, the writer's own source, read directly this run (2026-09-29) · spine/portal_write.py