XAIGuard™by HWITSYS
Runtime Enforcement

Stop the prompt before it leaves.

Every prompt and response is scored, redacted or blocked inline — with a verdict you can explain to an auditor.

A prompt data stream passing through four sequential enforcement gates, the third gate glowing red and blocking flow
How it works

From prompt to verdict in four stages.

  1. 01

    CLASSIFY

    The prompt is scored across eight risk dimensions.

  2. 02

    VALIDATE

    Scores are checked for consistency and false-positive pressure.

  3. 03

    POLICY

    Your natural-language and preset policies are evaluated against the scores.

  4. 04

    VERDICT

    Allow, redact, or block. The decision and its reasoning are written to the audit chain.

Capabilities

What it does.

RT-01

Eight-dimension risk scoring

PII, injection, jailbreak, toxicity, exfiltration, IP, policy intent and excessive agency — scored per prompt, per response.

RT-02

Inline DLP & redaction

Detect and redact sensitive entities inline so work continues without the data ever leaving the perimeter.

RT-03

Natural-language policy authoring

Write policy in plain English. XAIGuard compiles it to enforceable scoring thresholds — no rules DSL to learn.

RT-04

Preset policy packs

Opinionated starting packs for PII, source code, regulated data and prompt injection. Turn on, tune, done.

RT-05

Bulk scanning

Scan historical prompt logs, datasets and prompt libraries to find what already leaked — before your auditor does.

RT-06

REST scanning API

An OpenAI-compatible endpoint and standalone scoring API. Govern any SDK call by changing one base URL.

In practice

One policy, one explainable verdict.

policy · verdictlive

policy > "Never allow customer account numbers or unredacted contracts to reach an external model"

verdict: "BLOCK"

reason: "Customer account number detected in prompt"

matched_detectors: ["pii.account_number", "dlp.contract_text", "policy.nl-014"]

risk_scores: { pii: 0.98, exfiltration: 0.41, injection: 0.03 }

action: "block_and_alert"

latency_ms: 0.8

audit_hash: "ed25519:9f2c…a41e" — appended to tamper-evident chain

Framework mapping
OWASP LLM01 Prompt InjectionOWASP LLM02 Sensitive Information DisclosureOWASP LLM06 Excessive AgencyNIST AI RMF MEASURENIST AI RMF MANAGEMITRE ATLAS

Every verdict maps to the controls your frameworks already name — evidence, not assertion.

Related

Works together.

Deployment

PAC file, Browser Shield or API proxy — enforce from wherever your users already are.

Explore

Agent & MCP Governance

The same inline verdicts applied to autonomous agents and every MCP server they call.

Explore

Compliance & Evidence

Every allow, redact and block lands in a signed audit chain your auditor can verify.

Explore
Next step

See it run against your own prompts.

Request a demoTalk to security engineering