Claude-For-Mayor
Jerusalem elects a mayor on promises about schools, buses, policing, cleanliness and housing. A municipality's actual power over each of those ranges from near-total to essentially none — and the ranking is not the one anyone campaigns on. This repository is an experiment in building a mayoral platform properly: researched, costed, honest about its trade-offs, and refusing to promise anything the office cannot deliver.
Claude is not running for mayor of Jerusalem, is not eligible to, and does not live here. It is the sibling of MK-Claude, which does the same thing at national level — with one stage added, and that stage is the whole point.
The rule this project is built on
Most of what residents blame City Hall for is not City Hall's to fix. A good half of what City Hall could fix tomorrow, nobody asks it to. Working out which is which is not a preliminary to the policy work — it is the policy work.
No proposal is published without a named instrument and a named holder of it.Every lever is traced to the provision that creates it and classified: Athe municipality decides and funds it · B the municipality decides and a ministry must approve · C delivery is municipal but policy or funding is national · D not the municipality's at all.
Class D is legitimate as advocacy, with a named counterpart and a real route — never as a promise. And the mirror-image finding matters more: a power the municipality has held for thirty years and never used is the most valuable thing this process can turn up.The powers doctrine in full.
The platform, as it stands
Eight areas, all at stage one. They were opened on 10 August 2026 by a single dictated account of ten years living here — and under this project's own rules, not one of them may propose anything until its jurisdiction map exists. Eight areas and no jurisdiction maps is the wrong ratio, and it is the next thing to fix. Whose account that is, and what it cannot see.
Renting
The city bills, registers and communicates as though every household owns its home. The arnona arrives in the tenant’s name, cannot be paid monthly, and the risk when a tenancy ends mid-cycle lands on the person who owns nothing.
Stage 1 · testimony captured, jurisdiction gate not yet runHousing supply
A city consistently ranked among the poorest in Israel, adding high-rise stock built and marketed as luxury. How much of that unit mix a municipality actually controls is the open question, and it is not yet answered.
Stage 1 · testimony captured, jurisdiction gate not yet runConstruction disruption
Light rail works and tower construction running at once, main roads that simply end, no signage and no central place to find out what is closed. Building the network is defensible; the information vacuum around it may not be.
Stage 1 · two testimonies captured, jurisdiction gate not yet runConstruction noise
Deep-foundation excavation measured by a resident at 110 dB, with no published permit conditions, no duration cap anyone can point to, and no independent measurement. The objection is not the noise — it is that you cannot find out when it stops.
Stage 1 · testimony captured, jurisdiction gate not yet runSmall businesses
Traders closing under prolonged works with no compensation offered, against a collection machine described as extremely rigorous. Whether the city could relieve them, and whether any Israeli municipality ever has, is the question.
Stage 1 · testimony captured, jurisdiction gate not yet runEnvironmental quality
City-wide horn use and air quality below what a capital ought to have — two conditions everyone lives inside and nobody campaigns on. Also the area most likely to come back class D, which is exactly why it gets mapped before it gets promised.
Stage 1 · two-sentence testimony, jurisdiction gate not yet runCapital priorities
Tourism and prestige capital against neighbourhood infrastructure — a walkway to the Old City the author has never used in ten years here. It is an arithmetic claim, and it stands or falls on whether that money was ever the city’s to move.
Stage 1 · testimony captured, jurisdiction gate not yet runPublic transport
Frequency that fell away, a diverted route nobody could parse, no shade while you wait, and a fare-enforcement regime built for a two-dollar ticket. A mayor sets neither the timetable nor the fare — so the question is what the city does hold inside a nationally franchised service. Starting with the shelter over the stop.
Stage 1 · two testimonies captured, jurisdiction gate not yet run
Every one of those eight was opened by something going wrong, which is how the pipeline works and also its most predictable failure mode: a platform assembled entirely from grievances ends up describing a city nobody would choose to live in, and then proposes to govern it. The same account says what is good here — the pace, the familiarity, the institutions, and why the city should stop betting its future on events it does not control. What is good about the city.
Eight areas, fewer than eight problems
Read as one body of material rather than area by area, the account keeps describing the same few things. The strongest of them: the recoverable failure is usuallyinformational. Very rarely is the complaint that the works exist, that the towers are going up, or that the buses run — it is that nobody will say what is happening or when it stops.
That is not a rhetorical point. Almost every physical grievance here runs into a jurisdiction problem: the works are probably not the city's to run, the fare is not its to set, landlord–tenant law is national. Publishing what you already know is the act least likely to have that problem — nobody's confirmation is needed to put a map on a website or to answer a question.
Four more mechanisms sit alongside it, along with the counter-thesis that cuts against all of them — the city is doing the hard thing, and building a transit network is not a project anyone can make pleasant. None of it is verified, none of it proposes anything, and it is expected to be wrong in at least one part. The emerging spine · problems on file with no testimony behind them.
How a plank gets written
Four stages and a gate, each with an artifact and a bar it must clear before the next one starts. Nothing skips ahead.
- TestimonyA dated, cleaned first-person account. Evidence of experience — never of prevalence, and at city scale that distinction bites. It also yields a list of attributions to test: who the resident thinks is responsible. Read them.
- Jurisdiction — the gateEvery lever traced to a cited instrument and classified A/B/C/D, with a dedicated hunt for powers the city holds and does not use. The paper cannot be drafted until this exists.
- Evidence baseSix documents: the quantified problem by neighbourhood, the jurisdiction map, what is on the books today, comparator cities including at least one Israeli municipality and one instructive failure, a sourced statistical annex, and what City Hall has already tried and dropped.
- Policy paperA think-tank product. Every recommendation carries a class letter, a mechanism, a named budget line, a realistic route, and the failure mode it is designed against — plus a steelmanned trade-offs section naming who loses.
- Program for the cityThe planks rolled up and sorted by deliverability before theme, checked for department capacity, fiscal coherence and whether the council arithmetic actually delivers it. The pipeline in full.
What the ground looks like
The context the platform is written against, before a word of policy is drafted.
- 33.7%
Turnout in the February 2024 Jerusalem municipal election. Two-thirds of the electorate did not vote, in a mayoral contest decided 81–19. A platform aimed at people who already vote is aimed at a third of the city.
- 2 / 31
Council seats held by the mayor's own list. Moshe Lion won 81% of the mayoral vote and commands two seats of thirty-one; authority here is coalition authority, and any proposal needing a vote needs an answer to that.
- 16 / 31
Seats held by the Haredi lists between them — a bare majority on their own. There is no council majority that excludes them, which is a constraint on the platform rather than a complaint about it.
- 0
Powers verified so far. The standing powers primer is currently a list of hypotheses and a research plan, and it says so at the top. Under our own evidence rules none of it may be cited yet.
What this is not
Not a real candidacy. Not an endorsement of any faction on the council. Not a claim that AI should make policy — the human sets the values and the priorities; the AI does research, drafting and stress-testing, and gets told when it is wrong.
Think of it as civic engagement by other means: instead of complaining about the city, writing down — rigorously, and with the powers actually checked — what a platform worth voting for would contain. Corrections, better data and counterarguments are welcome, especially with sources. Open an issue.