Three tools · one ecosystem · no operator

Affected people.
Public data.
Inside sources.

Three open tools working on the same problem from different directions: making opacity and arbitrariness in the public sphere visible — and giving individuals concrete ways to act.

Evidentia, Offener Kataster and WARTE are independent. Each runs without the others. Together they form an infrastructure of non-forgetting: what disappears individually stays visible in aggregate. No central instance, no operator power, no kill switch. AGPL, federated, forkable. Read on if you want to know more.

I · Affected people

Evidentia

"I am currently experiencing
a legal problem."

Browser tool that guides you through any kind of legal problem — from police violence through bullying, discrimination, housing disputes, administrative arbitrariness to online fraud. Produces legally valid documentation, PDF records, submissions to authorities. The case stays local in the browser; only an anonymized fingerprint can optionally be linked to other tools.

78 situation types
110+ institution registers
Sensitivity tiers (4)
Youth protection
30-day quarantine
Single HTML, 333 KB

Status: Production-ready v3.5

II · Public data

Offener Kataster

"What stands out in the open,
but no one sees it?"

Backend system that systematically crawls public Austrian and EU data sources — RIS, Ediktsdatei, Parliament, procurement procedures, company register, WiEReG, lobby register, Ombudsman, Court of Audit reports, municipal minutes, CJEU, ECtHR. Detects patterns and contradictions across sources, rules, practices. All statements are hypotheses, not verdicts.

40+ data sources
12 domains
20+ pattern detectors
Contradiction engine
Protection-worthy module
PostgreSQL + pgvector

Status: Concept + ref impl v0.1

III · Inside sources

WARTE

"I see something,
that no one is meant to see."

Anonymous whistleblower service on Tor. Client-side encryption, key distribution via Shamir secret sharing to 5 independent recipients (3-of-5 threshold). No operator can decrypt the identity or the content — mathematically, not by policy. Metadata stripping, stylometric sanitizer, delay jitter against correlation attacks.

Onion service only
AES-GCM + Shamir(3,5)
6 threat vectors addressed
No logging, no plaintext
Reproducible builds
Tails ceremony

Status: Concept + protocol spec

How they connect

A shared language:
the fingerprint.

All three tools can produce the same anonymous data type: a cryptographic fingerprint containing institution hashes, time windows and situation tags — but no personal data. This lets pattern clusters be detected across tools without any exchange of identities.

━━━━ EVIDENTIA ━━━━ ━━━━ KATASTER ━━━━ ━━━━ WARTE ━━━━ Citizen documents Crawler combs through Insider reports anonymously Single case in the browser public sources via Tor/onion Produces: Produces: Produces: structure_hash entity_hashes institution_hashes entity_hashes[] event_ids (from report content) sensitivity_tier pattern_hypothesis recipient_set_id quarantine_until confidence_level token_hash (one-way) │ │ │ └──────────────┬───────────────┴────────────────────────────┘ │ ▼ [ matching pool ] (any instance, AGPL, several operable) Brings together what belongs together: · Same institution in 3 Evidentia FPs → cluster · Kataster pattern for this institution → reinforced · 1 WARTE report matches the cluster → confirmed │ ┌──────────┴───────────────────────────────┐ ▼ ▼ ▼ Back-signal to Dashboard Alert to NGOs Evidentia users: shows aggregated (Klagsverband, VKI, "The same office pattern publicly ZARA, Volksanwalt) was reported 4×" (no PII) when threshold reached NO ONE learns identities. Only institutions, frequencies, structures become visible.

In direct comparison

When do you use which?

Evidentia

Offener Kataster

WARTE

Who?

Affected citizens, personally involved

Journalists, NGOs, research

Staff, officials, witnesses

What?

Document your own case, exercise your rights

Structural patterns from public data

Report internal wrongdoing anonymously

How?

Browser tool, offline-capable

Backend crawler + dashboard

Tor hidden service

Risk?

Low — data stays local

Low — public data only

High — identity protection critical

Output

PDF record, submissions, fingerprint

Dashboard hypotheses, data export

Encrypted share to recipients

Status

Production-ready v3.5

Concept + ref impl

Concept + protocol

Prerequisite

Browser, nothing else

Server + PostgreSQL

Tor Browser + recipient set

Shared rules

What all three tools uphold.

∑-01 · AGPL open source

Anyone can fork, audit, self-host. No proprietarization.

∑-02 · federatable

Other organizations can run their own instances. Censorship resistance through decentralization.

∑-03 · no outing of persons

Natural persons are referenced only in a public role. No online-pillory logic.

∑-04 · minors never

Zero tolerance across all systems. DB constraints, UI warnings, manual review.

∑-05 · protection-worthy only structurally

Guardianship, detention, asylum, care: only aggregated, never individual data.

∑-06 · hypothesis before verdict

No findings, only pattern hints. Language discipline enforced in the code.

∑-07 · no commercialization

Not for scoring, HR checks, credit rating. Civic infrastructure, not a product economy.

∑-08 · right to rectification

Anyone appearing in the data can object. Errors are flagged, not silently deleted.

Implementation status

An honest self-assessment.

EVIDENTIA v3.5 production-ready · browser tool finished · Single HTML 333 KB, offline-capable, PWA installation possible · 78 situation types covered, 110+ institution registers · Sensitivity tiers, youth protection, cool-off quarantine · Fingerprint import with network graph · Diagnostic self-test tool Open: multilingual legal info, QR-code import OFFENER KATASTER Concept + reference implementation · Python/FastAPI backend, structured · PostgreSQL schema with pgvector, constraints · 3 reference crawlers (RIS, Ediktsdatei, Parliament) working · Pattern engine with 7 detectors + rule-02 validator · Contradiction engine designed · Protection-worthy module specified Open: 9 further crawlers, land-register access, municipal OCR WARTE Concept + protocol spec + prototype client · Threat model worked through (6 vectors) · Shamir secret sharing implemented and tested · FastAPI ingress server with hardening configuration · Tor deployment guide + ceremony protocol · Prototype UI with client-side crypto running Open: live deployment, recipient set under contract, security audit SHARED matching pool · the connecting node · Specification available · Anyone can run an instance (none central) · Legal review still pending Open: find concrete instance operators

What is realistically next

No longer code, but people.

The tools have reached the point where developer work alone no longer helps. What is needed now: lawyers for GDPR and media law, journalists for pattern validation, NGOs as recipients and brokers, funders for sustainable infrastructure.

LAW

GDPR legal opinion, media-law review, right-to-information strategies, possibly outreach to the data-protection authority

MEDIA

Investigative newsrooms as validation partners, journalists as test users of the Kataster API

NGO PARTNERS

Klagsverband, VKI, ZARA, Ombudsman, victim-protection bodies as brokers and WARTE recipients

FUNDING

netidee, NLnet, OTF, EU Direct Managed, NGI Zero — public funding programs for civil-society infrastructure

Affected people. Public data. Inside sources.