About Elect Righteous
Elect Righteous gathers the candidate information Hays and Kansas voters would otherwise have to hunt across filings, meeting records, campaign pages, finance reports, local coverage, and public online activity.
Mission
Local and state elections ask voters to make serious decisions with limited time, and the useful facts are scattered. This site puts those facts in one organized place so ordinary voters can compare candidates without spending hundreds of hours searching the internet.
The role is simple: make information easier to find, keep sources close, encourage people to vote, and give readers room to weigh candidates according to conscience, Scripture, and the values they believe should guide public office.
Elect Righteous does not endorse candidates or tell voters what conclusion to reach. It organizes public evidence, names source limits, welcomes documented corrections, and lets the reader judge.
Research engine
The research process is built from specialized agent prompts and large-language-model passes. Each pass has a defined job, and each output is checked against the files and sources that came before it.
Roster agents check Secretary of State filings, FEC records, KPDC records, county documents, city pages, school-board records, and election calendars. Filed candidates, current officials, off-cycle officeholders, administrators, and ballot measures are kept in separate categories.
Evidence-mining agents read the candidate folders, old research dumps, social harvests, reports, race files, source trails, and rendered data. Facts are pulled into evidence matrices before public prose is written.
Source-tier agents separate official records, candidate-controlled statements, reliable reporting, social evidence, and internal memory. Social observations such as posts, likes, follows, and comments are labeled as observations unless the candidate directly made the statement.
Each profile uses the same 14 issues. When no source-backed statement or action is found, the profile says that plainly instead of filling the gap with party assumptions.
The prompt files direct the workflow to notice truthfulness, stewardship, bribery, care for vulnerable people, public faith references, and consistency. That lens shapes what gets checked, while the public writing stays reportorial: no endorsements, no faith-based policy guesses, and no partisan shortcuts.
Editorial, legal-accuracy, symmetry, source-health, mobile-display, public-copy, build, GitHub Pages, and Cloudflare checks all sit at the end of the process before changes are treated as shipped.
The workflow is built around specialized prompt files and agent roles, not one generic summary request. Each role has a narrow job: roster audit, evidence extraction, source validation, social analysis, issue building, profile writing, review, validation, or deployment.
The system uses custom large-language-model workflows for extraction, classification, and writing support, but the durable record lives on disk: evidence matrices, source audits, issue matrices, social matrices, validation reports, and handoffs that the next run can inspect.
The agents are instructed to apply the same source order, issue list, caveats, and symmetry test to every person. The goal is not to erase worldview; it is to keep worldview from becoming party favoritism or unsourced inference.
Scale
The same core structure is used across candidates: identity and ballot status, the 14-issue matrix, public actions, social/online observations, donor and funding notes, worship/community references when public, source trails, and corrections.
