A local-first decision-governance system for preserving the reasoning, evidence, assumptions, risks, alternatives, and review history behind consequential decisions.
Try the live demo · See how scoring works · Architecture · Contribute
SOURCE → EVIDENCE → CLAIM → DECISION (with alternatives, assumptions, constraints, and risks)
↘ COUNTEREVIDENCE
Animated conceptual tour of the documented evidence model; this is not a fabricated live screen recording.
Validation Ledger is an MIT-licensed, local-first workspace for product managers, researchers, and technical leaders who need to preserve the chain from raw material to an explicit decision. While product discovery is a primary use case, it serves any domain requiring traceable decision governance.
- Evidence stays traceable. Quotes and evidence signals remain linked to the source material behind them.
- Counterevidence stays visible. Contradictory evidence is scored separately instead of disappearing inside an optimistic summary.
- The core product is local-first. Project data lives in IndexedDB in the current browser profile; there is no required application backend.
- AI is optional, not authoritative. Gemini extraction only runs when the user explicitly initiates it, and suggested evidence still requires human review.
- The scoring model is inspectable. Support strength is derived from explicit factors rather than a hidden model judgment.
Feedback wanted: where does this evidence model create false confidence, miss an important kind of discovery signal, or handle contradiction poorly? If the approach is useful, starring the repository helps other product teams discover it.
Built by David Turner · atomicdjt
The workspace includes project and claim and assumption management, a source library, source-level evidence extraction, an evidence matrix, decision tracking, reporting, backup controls, and guided demo data.
Validation Ledger is designed for people who need to synthesize qualitative customer evidence into defensible product decisions rather than relying on gut feeling, disconnected notes, or opaque AI summaries.
Typical workflows include:
- customer-interview synthesis;
- claim tracking;
- product-discovery evidence review;
- explicit decision records;
- preservation of contradictory signals;
- research handoff and auditability.
- React 19, React Router 7, TypeScript 6, and Vite 8
- Tailwind CSS 4 with a custom token-based design system
- Dexie 4 and IndexedDB for browser-local persistence
- Zustand for small persisted interface preferences
- Google Gemini as an optional, user-configured extraction assistant
- Vitest and Oxlint for deterministic tests and static analysis
- Vercel for the static production deployment
Requirements: Node.js 24 and npm 11.
git clone https://github.com/atomicdjt/validation-ledger.git
cd validation-ledger
npm ci
npm run devOpen the local URL shown by Vite. No environment variable is required for the core product. To use AI extraction, add a restricted Gemini API key in Settings; never commit a real key.
Telemetry is off unless VITE_POSTHOG_KEY is configured at build time (and VITE_POSTHOG_HOST is optional). It records only these coarse completion events: application load; project, source, claim, manual-evidence, and decision creation; source-note save; and backup export/import. Allowed properties are limited to type, enum, boolean, and link-count metadata.
It never sends source notes, evidence or claim text, user-provided identifiers, URLs, backup content, local database state, API keys, or user identity. Application events use finite structural values only. PostHog still receives the public project token in the transport request and may attach anonymous SDK identifiers and library/timestamp envelope fields required for ingestion; browser, device, URL/referrer, IP-derived location, and session-recording fields are disabled or blacklisted. Missing or failed telemetry configuration cannot affect core product behavior.
To verify ingestion, add the PostHog public project key as VITE_POSTHOG_KEY in Vercel, redeploy, complete a single non-sensitive workflow, and inspect PostHog Live Events for an allowlisted event name. Add future events only through src/services/analytics.ts with a narrow typed property contract and a corresponding test.
npm test
npm run lint
npm run build
npm auditThe scoring suite covers source independence, segment diversity, behavioral evidence, direct citations, neutral/missing relationships, counterevidence semantics, and source-text revalidation. GitHub Actions repeats unit tests, lint, production build, dependency audit, Chromium workflow, and automated accessibility checks on every push and pull request.
The IndexedDB database is named ValidationLedgerDatabase. Data is tied to the current browser profile and site origin. Use Settings → Export backup regularly; clearing browser storage can remove local records. Current-format backup import validates the supported schema and verifies hash-linked revision chains before replacing current data in one transaction. Format-1 backups can be restored for data preservation, but they predate revision history and are explicitly marked as historically unverifiable.
The live demo does not receive or retain application data on a server. Optional AI extraction sends the selected text and request directly from the browser to Google's API using the key supplied by the user.
Validation Ledger reports support and counterevidence separately. Supporting strength combines:
- independent supporting sources, capped at 60 points;
- segment diversity, capped at 15 points;
- behavioral or willingness-to-pay evidence, worth 15 points;
- direct source citations, capped at 10 points.
Counterevidence receives its own independently capped score. Credible support plus contradiction is labeled mixed; contradiction without support is contradicted. This keeps positive evidence from erasing material counterevidence.
Only excerpts that match the saved source exactly or through conservative whitespace/quotation normalization may be treated as direct evidence. Missing or unverified provenance is displayed as an inference, never as a quote. The score is an aid to judgment, not proof of market demand. AI-generated suggestions also require human review.
The technical evidence for the project is documented rather than implied:
- Architecture and system boundaries
- Governance Model
- Release checks and current limitations
- Security and responsible key handling
- Contribution workflow
The scoring system is intentionally heuristic. It structures judgment; it does not prove product-market fit, causal demand, source truthfulness, or research quality.
BuildWorld AI · WeaveStudio · GitHub profile · Full portfolio
Validation Ledger is open source under the MIT License. Copyright © 2026 atomicdjt.