Griha: the chore ledger where the fairness argument is auditable, and TabPFN runs in your tab

Description and use

Griha is an offline‑first chore ledger that lets users claim tasks, view detailed fairness calculations, and export verifiable data. TabPFN runs locally in the browser, ensuring privacy and offline capability.

Problem

In shared homes, a de facto project manager tracks chores but lacks transparent, auditable evidence of fairness, leading to unresolved disputes.

Outcome

Griha offers a transparent, offline‑capable chore ledger with auditable fairness metrics and local ML predictions, eliminating the need for a hosted backend and enabling verifiable dispute resolution.

Workflow

• 1. User claims a chore by selecting it on the board; the claim is recorded without double‑asking.
• 2. The app calculates a fairness score using computeFairness, showing each term’s contribution.
• 3. Users can export the ledger in various formats, including raw JSON with an audit chain for verification.
• 4. The TabPFN model predicts likelihood of flaking, running entirely in the browser via WebTabPFN.
• 5. Users can review the audit chain by replaying seals to confirm integrity.

Evidence

Griha is a chore ledger for that person. Three things, and only three things, it does: Claim a chore without asking twice. See the arithmetic, not a verdict. Export something checkable.

Source

It runs TabPFN v2 — Prior Labs' open weights — through the WebTabPFN runtime, inside the tab. Not a hosted endpoint.

Source

109 unit and integration tests. 91 live checks. 26 browser journeys. 20 rows for TabPFN. Green CI on both jobs.

Source