PII Integrity · Deterministic Infrastructure · Secure Protocol
The protocol layer between intent and execution in AI systems.
Every signal examined. Every verdict signed. Nothing passes without record.
This repository exists for transparency and contribution — not as a deployable alternative to the hosted service.
| What's in this repo | What's not in this repo |
|---|---|
Authentication middleware (/middleware) |
IntegrityGuard detection engine |
Stripe webhook handler (/api/webhooks) |
PrivacyGuard NLP classification layer |
Supabase schema & migrations (/supabase) |
Signature database (22 injection vectors) |
| API interface contracts & response shapes | Model weights and training data |
Landing page & documentation (index.html) |
Audit record signing infrastructure |
Cloning this repository does not give you access to the Zentric processing service. The detection engine that inspects prompts, detects PII, and generates signed audit reports runs on Zentric's infrastructure and requires an active license.
Because trust is infrastructure. You should be able to verify how authentication works, how your API key is validated, and how subscription state is checked before your requests reach the engine. We believe in auditability at every layer — including our own enforcement code.
We accept contributions to the middleware, webhook handler, and Supabase schema. Open a PR or file an issue. For security-related contributions, see the Security section.
| Tier | Price | Requests | Start |
|---|---|---|---|
| Free Trial | Free | 2,000 requests | Get API key → |
| Growth | $499/mo | 100,000 req/mo | Start Growth → |
| Enterprise | $2,500/mo | Unlimited | Start Enterprise → |
Zentric Protocol is an infrastructure integrity layer for AI systems. It sits between your application and your LLM, examining every signal — prompts, responses, user inputs — and returning a cryptographically-signed verdict before execution continues.
It is not a filter. It does not guess. It applies deterministic rules across a standardized pipeline and returns a structured, auditable JSON report for every request.
Input Signal
│
▼
┌─────────────────────────────────────────┐
│ ZENTRIC PROTOCOL │
│ │
│ ┌─────────────┐ ┌─────────────────┐ │
│ │IntegrityGuard│→│ PrivacyGuard │ │
│ │ 22 injection │ │ 17 PII types │ │
│ │ signatures │ │ 7 languages │ │
│ └─────────────┘ └────────┬────────┘ │
│ ▼ │
│ ┌──────────────┐ │
│ │ ZentricReport│ │
│ │ UUID+SHA-256 │ │
│ │ GDPR Art.30 │ │
│ └──────────────┘ │
└─────────────────────────────────────────┘
│
▼
Verdict + Certificate → Your System
Extracted from Zentric Integrity Report v1.0 — 1,000,000 simulations across all supported attack vectors and entity types.
| Attack Vector | Simulations | Detected | Precision |
|---|---|---|---|
| Prompt Injection (EN) | 187,430 | 187,012 | 99.78% |
| Prompt Injection (ES/FR/DE) | 134,210 | 133,401 | 99.40% |
| Base64 / Token Smuggling | 48,900 | 48,761 | 99.72% |
| Jailbreak multi-vector | 67,340 | 66,988 | 99.48% |
| Fake SYSTEM override | 39,120 | 39,087 | 99.92% |
| Role redefinition | 52,000 | 51,743 | 99.51% |
| Total | 529,000 | 528,992 | 99.62% |
Full methodology and raw data available on request: core@zentricprotocol.com
The 1,000,000-simulation corpus was constructed from four sources: (1) published prompt injection research datasets (PINT Benchmark, PromptBench, garak); (2) adversarial samples hand-authored across 22 attack categories to stress-test edge cases not present in public datasets; (3) synthetically mutated variants of known attack patterns — character substitution, whitespace injection, Unicode normalization attacks, and mixed-language payloads — to measure robustness against obfuscation; and (4) benign control samples drawn from production-representative traffic to verify precision does not degrade under normal use. The final split is approximately 53% attack samples and 47% benign controls. Simulations were run deterministically: the same corpus against a frozen signature set, with no retraining or tuning between runs.
What this benchmark does not cover: adversarial inputs specifically constructed to defeat these 22 signatures after reading this repository (signatures are partially public, which is a known trade-off of transparency); semantic prompt injections that do not match any current signature pattern and rely entirely on model misinterpretation; non-English languages beyond the seven supported (EN, ES, FR, DE, IT, PT, NL); and multi-turn conversation-level attacks where the injection is distributed across several messages. The precision figures above reflect deterministic pattern matching — not an ML classifier, not a generalization claim. Zentric is honest about what it detects and what it does not.
Detects prompt injection, jailbreak attempts, and instruction overrides before they reach your LLM.
- 22 catalogued injection signatures
- 7 supported languages (EN, ES, FR, DE, IT, PT, NL)
- Multilingual NLP classification layer
- Mean server-side processing: 23.4ms (network latency not included)
Identifies and anonymizes PII in prompts and responses. Regional standards treated as first-class entities.
- 17 PII entity types: SSN, NIF, CPF, CURP, IBAN, SWIFT, passport, email, phone, and more
- Regional pattern recognition (EU, US, LATAM)
- Anonymization operators: redact, mask, tokenize, pseudonymize
- Recall rate: 99.71% across 17 entity types
Every request that passes through the protocol generates a signed, immutable audit record.
{
"report_id": "zp_01HXYZ...",
"uuid": "f47ac10b-58cc-4372-a567-0e02b2c3d479",
"timestamp_utc": "2026-05-14T22:00:00.000Z",
"sha256": "e3b0c44298fc1c149afb...",
"verdict": "CLEARED",
"integrity": {
"injection_detected": false,
"signatures_matched": [],
"confidence": 0.9998
},
"privacy": {
"pii_detected": true,
"entities": [
{ "type": "EMAIL", "action": "REDACTED", "position": [42, 61] }
]
},
"compliance": {
"gdpr_art30": true,
"ccpa": true,
"eu_ai_act_s52": true
},
"latency_ms": 21.4
}curl -X POST https://api.zentricprotocol.com/v1/analyze \
-H "Authorization: Bearer zp_live_..." \
-H "Content-Type: application/json" \
-d '{
"input": "Your prompt or user input here",
"modules": ["integrity", "privacy"],
"options": {
"anonymize": true,
"language": "auto"
}
}'{
"status": "ok",
"verdict": "CLEARED",
"report": { ... },
"anonymized_input": "Your prompt or user input here",
"latency_ms": 23.1
}| Verdict | Description |
|---|---|
CLEARED |
Input passed all checks. Safe to forward to LLM. |
BLOCKED |
Injection or high-risk pattern detected. Reject. |
ANONYMIZED |
PII found and redacted. Anonymized input returned. |
REVIEW |
Low-confidence detection. Human review recommended. |
| Language | Status |
|---|---|
| Python | pip install zentricprotocol (coming Q3 2026) |
| Node.js | npm install @zentricprotocol/sdk (coming Q3 2026) |
| REST API | Available now |
Zentric Protocol is designed from the ground up for regulated AI deployments.
| Standard | Coverage |
|---|---|
| GDPR Art. 30 | Reproducible audit record per request — one component of an Art.30 documentation strategy |
| GDPR Art. 25 | Privacy by design — anonymization as default |
| CCPA §1798.100 | Consumer data identification and processing record |
| EU AI Act §52 | Transparency obligations resolved at infrastructure level |
| SOC 2 Type II | Audit trail and access controls (in progress) |
| Tier | Price | Requests | Use Case |
|---|---|---|---|
| Growth | $499/mo | 100,000 req/mo | AI-forward companies in production |
| Enterprise | $2,500/mo | Unlimited | Regulated industries, EU data residency, dedicated SLA |
→ Start Growth · → Start Enterprise · → Contact for custom
Deterministic. The same input always produces the same verdict. No probabilistic black boxes in the critical path.
Stateless. The protocol does not store your data. Each request is processed and returned. The audit record is yours.
Composable. Deploy the full stack, a single guard, or wire only the audit layer into existing infrastructure.
Auditable. Every verdict is signed with SHA-256, timestamped in UTC, and assigned a UUID. Your compliance team will thank you.
We take the security of this protocol seriously. If you discover a vulnerability, please report it responsibly.
- Email: core@zentricprotocol.com
- Subject:
[SECURITY] <brief description> - Response SLA: 48 hours acknowledgement, 7 days resolution target
We do not operate a public bug bounty program at this time. Responsible disclosure is acknowledged in our changelog.
- IntegrityGuard v1.0 — 22 signatures, 7 languages
- PrivacyGuard v1.0 — 17 PII types, EU/US/LATAM
- ZentricReport v1.0 — SHA-256, UUID, GDPR Art.30
- REST API (production)
- Python SDK — Q3 2026
- Node.js SDK — Q3 2026
- Streaming support (SSE) — Q3 2026
- Webhook callbacks — Q4 2026
- SOC 2 Type II certification — Q4 2026
- Self-hosted deployment option — 2027
| Channel | |
|---|---|
| General | core@zentricprotocol.com |
| Enterprise | core@zentricprotocol.com |
| Security | core@zentricprotocol.com |
| X / Twitter | @ZentricProtocol |
| Zentric Protocol |
Zentric Protocol · Infrastructure Integrity for the AI Era
zentricprotocol.com · © ZP MMXXVI · v1.0.0
Built for CTOs who know that trust is infrastructure, not a feature.