About the site
One body of information, based on facts, about every list running in the election to the 26th Knesset - so that you can decide for yourself.
Why we built this
Before the election we tried to understand what the parties are actually offering, and got stuck. Every party says whatever suits it about itself, the platforms are long or do not exist, and what is said in the plenum, in committees and on social media is scattered across dozens of places. In the end we found ourselves voting on gut feeling.
So we collected the information ourselves. This site is the result: what every list says, what it has done in the Knesset, and what it stays silent on - in its own words, with a link to the source.
What you will find here
- The same questions for everyone. Every list is described using exactly the same set of topics, so they can be compared.
- Their words, not ours. Every claim rests on an exact quote and a link to where it was said.
- Silence is information too. If a list has said nothing about a topic, this is stated explicitly.
Where the information comes from
We searched everywhere a party, its Knesset members, its ministers and its candidates speak in their own name:
Platforms and party websites
What the party published itself. Where it exists, it comes before anything else.
Bills
Bills that members of the list submitted or signed, from the Knesset's legislation database.
Speeches in the plenum
What was said from the Knesset podium, from the official Knesset record.
Committee transcripts
What Knesset members said in committee debates, from the transcripts.
Ministers and government
Ministers' answers in the Knesset, and decisions and official announcements of ministries headed by people from the list.
Social media
Posts on the official accounts of the party and of the candidates - X, Facebook, Instagram and Telegram.
Interviews and media reports
Direct quotes from people on the list in the press, including the Arabic- and Russian-language press. Used only when the party itself has not published a position, and marked as such.
Media questionnaires
Answers the parties gave to questionnaires from media outlets, with the name of the outlet that asked.
Official records
The Central Elections Committee and the Registrar of Parties - for facts (who is on the list, in which place), never as evidence of a position.
A quote translated into Hebrew is marked "(תרגום)" ("translation"). The original, in its original language, is the evidence; the translation is ours.
How we worked
- Collecting. We ran AI agents 1,085 times, many of them in parallel, to scan within weeks an amount of material a small team could not have read in a year: thousands of pages of protocols, bills, speeches and posts.
- Checking. AI makes mistakes, makes things up and rewords. So no quote went up on the site just because a machine found it: every quote was checked by a person against the source - that the words are exact, that they were said by the person named, and that they were not taken out of context.
- Publishing by the rules. An exact quote and a link, or it does not go in. No summaries of ours in place of their words, no "according to estimates", no interpretation of intentions.
In numbers
- 14lists
- 1,596quotes, each with a link
- 1,097sources read
- Every quote checked by a person
- 5+ billiontokens processed by the AI models
- 36.7 millionwords in the saved source documents
- 3,365source documents saved
- 1,085AI agent runs
- 2,788quotes checked word for word against the saved source
- 3,469searches and downloads logged
How we counted
- tokens processed by the AI models: 5+ billion. The sum of the token usage fields (input, output, cache reads and cache writes) across all of the project's Claude Code work transcripts. Most tokens are the same context re-read by the model again and again; the number shows how much work the models did, not how much text exists.
- words the agents read: 16,490,249. Words in the tool results returned to the agents in the work transcripts.
- words in the saved source documents: 36,741,793. Words in the text files of the source documents saved in the research folder.
- source documents saved: 3,365. Documents saved in the research folder; a PDF and the text extracted from it count as one document.
- AI agent runs: 1,085. Runs of the agent tool in the work transcripts.
- actions (searches, reads, checks): 37,611. Tool calls (a search, a file read, a command run) in the work transcripts.
- searches and downloads logged: 3,469. Lines in the search logs (searches.yaml) of all the lists.
- quotes collected, each with a link and a date: 2,832. Lines in the quote files of the 14 lists.
- quotes checked word for word against the saved source: 2,788. Quotes put through the quote-checking tool in its run of 5 October 2026 on all 14 lists: the tool looked for each quote word for word in the saved copy of the source.
- bills saved from the Knesset records: 884. Distinct private member's bills saved from the Knesset records.
- Knesset protocols saved: 200. Knesset protocol files that were quoted from.
- written working rules: 97. Decision items in the project's decisions file.
Measured on 7 October 2026. The measurement can be re-run: tools/scale.mjs.
What is not here
- No polls.
- No recommendation on who to vote for. The site's quiz compares your answers with what each list said, and links to the quote.
- No comments or discussions.
- No funding from any party, candidate or political body.
A mistake? Something missing?
We worked carefully, and we may still have missed something. If you found an inaccurate quote, a broken source link or a missing position - write to us and we will fix it. Party representatives are welcome to contact us too: if you have published a position that does not appear here, send us a link.
Privacy
We count visits with Cloudflare Web Analytics, without cookies and without collecting personal details, only to know which pages are read. The forms on the site (report an error and contact us) send what you write to our email inbox, and none of it is stored on the site. Cloudflare's privacy policy.
Who we are
A small group of citizens who care that the election is held with the information open to everyone. We are not a party, not a campaign and not a media outlet, and we have no connection to any list.