Parser-only instruction found in non-visible HTML block.
Local-first scanner plus MCP guardrail
Catch hidden instructions before they reach your model.
Veridicus Scan inspects HTTPS URLs plus PDF, DOCX, and HTML files for hidden prompt-injection patterns, suspicious metadata, parser-visible drift, and risky redirect behavior. Premium also unlocks a local MCP mode for App Intents and agent workflows while the app is active.
scan.session
LOCAL
Veridicus Scan
Hidden instruction inspection report
Risk score
84
Directive-style language detected in a document metadata field.
Redirect blocked before the destination fetch continued in strict URL mode.
report.export = pdf,json
source.trust = direct
Explore
Start with the pages that explain the risk model, product boundaries, and real workflows.
If you are new to the site, these three pages are the quickest way to understand what Veridicus Scan is for, what it covers today, and where it fits in an AI-assisted workflow.
Why it matters
See the gap between what a person sees, what parsers normalize, and why hidden instructions matter before content reaches AI.
Coverage
Review the exact source types, hidden channels, report outputs, and partial-coverage notes the app actually supports today.
Use cases
Go from headline messaging into real scenarios for assistant safety, document review, evidence export, and MCP automation.
Start here
Use the site in three tracks: foundations, product reference, and agent workflows.
If you want the shortest path through the site, start with the explainers that define prompt injection, then move into the product pages for coverage, trust boundaries, reports, MCP, and OpenClaw workflows.
OpenClaw workflows
Put a local review step in front of OpenClaw.
If OpenClaw is going to fetch links, read files, or run tool-enabled workflows, review the intake first. Set up the agent, scan the risky material locally, and only then pass reviewed content into the workflow.
Setup
Guard
Automate
How to set up OpenClaw on macOS
Install the CLI, finish onboarding, pair a channel, and run the basic health checks before you wire in new workflows.
Use Veridicus Scan as the OpenClaw safety layer
Screen suspicious links, PDFs, DOCX files, and HTML locally before OpenClaw fetches, uploads, or reasons over them.
Add a local MCP review step
Keep the intake decision close to the agent boundary when you want repeatable local automation instead of blind ingestion.
Workflow
Review first. Then pass content to the model.
Choose the source, inspect the hidden signals, then review a scored report before handoff.
Choose the source
Start from an HTTPS URL or import a PDF, DOCX, or HTML file.
Inspect hidden signals
Review normalized text, hidden channels, metadata, and redirect behavior.
Export a usable report
Every scan ends with findings, guidance, coverage notes, and PDF or JSON export.
Local MCP workflows
Keep the scan, report, and approval step on-device.
Premium adds a local MCP bridge so an agent can scan, fetch reports, redact memory, and gate risky actions without sending the workflow to a remote daemon first.
Scan and report
Open a session, scan text or URLs, fetch the report, then export or close.
Memory controls
Use memory ingest, retrieval, and selective disclosure when the agent needs tighter runtime handling.
Plan and action gates
Scope tools, guard plans, and gate actions instead of treating the scan as a passive score.
Report-first trust
Finish every scan with a report you can use.
The report carries the score, band, findings, coverage state, and export controls that matter after the scan.
REPORT / FINDINGS / LOCAL EXPORT
source
https://example.com/help-center
finding
parser-only instruction found in hidden DOM
evidence
non-visible block contained directive-style control language
action
review before sending into an assistant or sharing into an AI workflow
Trust model
Local by default. Explicit when the network is involved.
Users should be able to tell what stays on-device, when a URL is fetched, and how exported evidence is handled.
On-device first
Analyze locally and generate the report on-device by default.
Clear fetch boundaries
Keep HTTPS-only input, redirect handling, and partial-coverage behavior explicit.
Evidence you can review
Make findings and coverage notes readable enough to support a real decision.
FAQ
Common questions about scanning, reports, and local MCP workflows.
What does Veridicus Scan inspect?
HTTPS URLs plus imported PDF, DOCX, and HTML files. The scan looks for hidden prompt-injection patterns, metadata anomalies, and parser-visible signals a person may miss.
How do URL scans handle redirects?
URL input is HTTPS-only. Strict mode blocks cross-URL redirects. Lenient mode follows HTTPS-safe redirects for broader compatibility.
What can I export after a scan?
You can export JSON or PDF. Evidence snippets are redacted by default unless you choose to include them.
What is the MCP feature?
Premium adds a local foreground MCP and App Intent bridge for session-based agent workflows. It supports Scan Text, Scan URL, Get Report, Export Report, plus Selective Disclosure, Guard Plan, and Gate Action while the app is active.
Why would a report mention partial coverage?
If a fetched response or imported file exceeds the scan budget, the report is marked partial and includes coverage notes so the result is not overstated.
Get started
Inspect the content before the model does.
Veridicus Scan is for people who want local inspection, readable evidence, and clear boundaries before content enters an AI workflow.