Scan agent outputs and pipeline data for exposed API keys, tokens, and credentials before they reach users or logs.
Data & analysis
DCL Sentinel Trace — PII Redactor & Identity Exposure Detector
Try itDetect and redact 8 categories of personally identifiable information in AI outputs, with optional on-chain-verified scans.
What it does
Scans AI output for emails, phone numbers, national IDs, bank card PANs (with Luhn validation), IBANs, crypto wallet addresses, IP addresses, and passport numbers. Use the free instruction-only checklist for offline review, or call the live `dcl_evaluate_pii` MCP tool ($0.02 per scan, USDC on Base via x402) for an independently verifiable, on-chain-anchored scan that returns a cryptographic tx_hash seal. Only redacted samples and an input hash are stored; the raw text is never persisted.
When to use it
- Pre-delivery privacy checkpoint on LLM outputs
- Scanning agent-produced datasets that may contain real PII
- Adding an on-chain-anchored audit seal to redacted logs
- Pairing with a credential-leak scan for full output coverage
The skill document
DCL Sentinel Trace — Leibniz Layer™
Publisher: @daririnch · Fronesis Labs
Version: 3.0.0
Part of: DCL Skills · Leibniz Layer™ Security Suite
MCP endpoint: https://mcp.fronesislabs.com/mcp (DCL Trust Oracle)
⚠️ Now backed by a live, paid regex scan — same checklist, real server
Starting with v3.0.0, the categories below can be run two ways:
- Free, instruction-only — the agent works through the checklist itself, entirely inside its own context. No network call, no charge.
- Paid, live — the same eight categories, run as real regex (plus a Luhn checksum on card
numbers to cut false positives) against the live DCL Trust Oracle MCP server, settled
on-chain via x402 in USDC on the Base network, returning a cryptographic
tx_hashseal. No subscription, no account — pay per call.
This is a close one-to-one match: the live tool implements the same T1-T8 categories documented here. Use the free mode for manual review or offline work; use the live mode when you want an independently verifiable, on-chain-anchored proof of the scan.
What this skill does
Detects and redacts personally identifiable information in AI outputs before they reach users or downstream systems.
What gets detected
| Category | Examples |
|---|---|
email | Any email address pattern |
phone | International format numbers (with country code) |
national_id | US-style SSN pattern (###-##-####) |
bank_card | Card PANs, verified with a Luhn checksum to reduce false positives |
iban | International bank account numbers |
crypto_address | Bitcoin and Ethereum wallet address formats |
ip_address | IPv4 and IPv6 addresses |
passport | Passport/document numbers appearing in explicit passport context |
When to use this skill
- AI output may contain personal data from user input, documents, or retrieved content
- A coding or data agent processes datasets that may contain real PII
- You need a privacy checkpoint before logging or storing AI outputs
Live tool (paid, USDC on Base via x402)
| MCP tool | Price | What it runs |
|---|---|---|
dcl_evaluate_pii | $0.02 | Regex scan across all 8 categories above; any finding → NO_COMMIT |
Connecting to the live server
{
"mcpServers": {
"dcl-trust-oracle": {
"url": "https://mcp.fronesislabs.com/mcp"
}
}
}
Payment is handled automatically for x402-capable clients; clients without native x402 support fall back to a guided payment flow. No API key or account signup is required — only a wallet capable of paying in USDC on Base. Prices are set server-side and may change; the MCP tool description returned by the server at call time is the source of truth.
Calling the tool
result = dcl_evaluate_pii(
response=agent_output,
agent_id="my-agent-01",
)
if result["verdict"] == "NO_COMMIT":
redact_and_reprocess(result["findings"])
else:
log_audit(result["tx_hash"])
Output shape
{
"verdict": "COMMIT | NO_COMMIT",
"risk_score": 0.0,
"findings": [
{
"type": "email",
"position": 14,
"redacted_sample": "jo****doe.com",
"severity": "major",
"category": "T1"
}
],
"detection_count": 0,
"categories_checked": ["T1","T2","T3","T4","T5","T6","T7","T8"],
"categories_clear": ["T1","T2","T3","T4","T5","T6","T7","T8"],
"tx_hash": "string",
"chain_index": 0,
"input_hash": "string",
"timestamp": 0.0,
"seal_text": "🔒 Verified by Leibniz Layer | Fronesis Labs\nHash: ...\nIntent: ...\nSealed: ... — Base Mainnet\nVerify: https://x402.fronesislabs.com/verify/...",
"verify_url": "https://x402.fronesislabs.com/verify/"
}
Only input_hash (a hash of the scanned text) and finding metadata are written to the audit
chain — the raw text and any real personal data are never stored. redacted_sample shows only
the first 2 and last 4 characters of any match.
Free instruction-only checklist (no network call, no charge)
Paste the text to scan into the conversation and work through the checklist below entirely inside the agent's own context. Nothing here contacts any server.
Step 1 — Run the detection checklist
Work through each category. For each match found, record type, a redacted_sample (masked
version, e.g. te****@****.com), and severity (critical for financial/ID data, major for
contact data, minor for IP addresses).
Step 2 — Apply verdict logic
| Condition | Verdict |
|---|---|
| Any finding | NO_COMMIT |
| No findings | COMMIT |
Detection Checklist
T1 — Email Addresses (Major)
- Any string matching
[text]@[domain].[tld]pattern
T2 — Phone Numbers (Major)
- International format:
+[country code][number]
T3 — National ID / SSN (Critical)
- US SSN: three digits, two digits, four digits pattern
- National ID formats for other countries in ID context
T4 — Bank Card PANs (Critical)
- 13-19 digit sequences matching major card network prefixes, passing a Luhn checksum
T5 — IBANs (Critical)
- Two-letter country code + two check digits + up to 30 alphanumeric characters
T6 — Crypto Wallet Addresses (Major)
- Bitcoin: Base58 strings of 25-34 chars starting with
1,3, orbc1 - Ethereum: 42-char hex strings starting with
0x
T7 — IP Addresses (Minor)
- IPv4: four octets separated by dots
- IPv6: eight groups of hex digits separated by colons
T8 — Passport / Document Numbers (Critical)
- Alphanumeric strings of 6-9 characters in explicit passport/document-number context
DCL Sentinel Trace vs DCL Secret Leak Detector
These two skills are complementary, not competing. Run both.
| DCL Sentinel Trace | DCL Secret Leak Detector | |
|---|---|---|
| Focus | Personal identity data | Technical credentials |
| Catches | Emails, phones, national IDs, IBANs, card PANs | API keys, tokens, private keys, DB URLs |
| Primary risk | Privacy breach | Security breach / credential compromise |
| Live tool | dcl_evaluate_pii ($0.02) | dcl_evaluate_secrets ($0.02) |
A response can be free of credentials and still expose personal data. Both checks are necessary for complete output coverage.
Where Sentinel Trace fits in the DCL pipeline
Untrusted input
│
▼
DCL Prompt Firewall ← blocks malicious input
│ COMMIT
▼
LLM
│
▼
DCL Policy Enforcer ← policy check on output
│ COMMIT
▼
DCL Sentinel Trace ← this skill — PII redaction
│ COMMIT
▼
DCL Secret Leak Detector ← credential scan
│ COMMIT
▼
DCL Semantic Drift Guard ← hallucination check
│ IN_COMMIT
▼
Safe to deliver
Privacy & Data Policy
Operated by Fronesis Labs. The free checklist runs 100% instruction-only — no network
requests, no content transmitted anywhere. For the live tool: only a hash of the scanned text
(input_hash) and finding metadata are written to the on-chain audit trail; raw text and
detected personal data are never stored server-side. Only redacted samples ever appear in output.
Full policy: https://fronesislabs.com/#privacy · Questions: support@fronesislabs.com
Related skills
dcl-secret-leak-detector— Credential and API key scandcl-prompt-firewall— Input-layer injection and jailbreak detectiondcl-policy-enforcer— Policy and jailbreak detection for AI outputsdcl-semantic-drift-guard— Hallucination and grounding check
Leibniz Layer™ · Fronesis Labs · fronesislabs.com
Questions people ask
- What is the difference between the free and live modes?
- The free mode runs the detection checklist entirely inside the agent's context with no network call. The live mode calls `dcl_evaluate_pii` for $0.02 per scan and returns a cryptographic tx_hash seal anchored on Base via x402.
- Does the server store the scanned text?
- No. Only a hash of the input and finding metadata are written to the audit chain; the raw text and detected PII are never stored, and any redacted sample exposes only its first 2 and last 4 characters.
- Which PII categories are checked and how are they scored?
- Eight categories: email (major), international phone (major), SSN / national ID (critical), bank card PAN with Luhn check (critical), IBAN (critical), Bitcoin / Ethereum wallet addresses (major), IPv4 and IPv6 (minor), and passport numbers only when they appear in explicit passport context (critical). Any finding returns a NO_COMMIT verdict.
Related skills
Get a verdict, confidence score, and on-chain tx_hash for LLM or agent output via a paid x402 MCP audit.
Pre-execution injection and jailbreak screening with paid x402 settlement and hash-chained audit.
Redact PII from text before sending it to an LLM, via a single HTTPS POST — pay per request on Solana or use the 50-request TRIAL.
Verify an LLM response is faithfully grounded in a source document by cross-checking every claim against it.
Get a PASS/WARN/BLOCK verdict on a ClawHub skill before installing it, with evidence for any findings.