Evidence-first modular intelligence

Turn public signals into reviewable intelligence workflows.

LumenRadar organizes source collection, evidence, dialectic checks, reports, review, cost controls, and operator workflows into a modular intelligence platform. Public pages show the safe product shape; stronger tools remain internal and review-gated.

View sample outputs Explore platform centers

Boundary

This production public site does not enable customer delivery, public MCP, real payments, or real customer data.

Current public boundary

Evidence chainsample evidence
Human reviewrequired
Public API/MCPnot open
Customer deliverydisabled

What the site can show

Modulessample registry
Reportsinternal drafts
Pricingcost-aware draft
Policysafety boundary

Platform map

One system, separated public and internal surfaces

The public site describes the product shape while the owner cockpit keeps evidence, review, cost, and tool control internal.

MCP Weapon Layer

public preview

Tool registration, risk state, and capability labels are shown as local preview inventory.

Evidence Chain

public preview

Signals become evidence records with quality, limitations, and audit visibility.

Reviewable Reports

public preview

Draft reports stay internal until review, evidence, and future policy gates allow release.

Human Review

public preview

Operators can approve internally, request revision, or reject customer delivery.

Admin Cockpit

public preview

The owner cockpit connects modules, evidence, review, finance, tools, and status pages behind protection.

Cost-aware Intelligence

public preview

Sample usage logs drive local cost summaries without real billing or payments.

Upgrade-ready Crawlers

public preview

Crawler capability is classified by risk level; high-risk live scraping stays disabled.

Internal Strong Port

public preview

Powerful operations remain private, review-gated, and unavailable from public pages.

Public Safe Port

public preview

The public site explains the platform without exposing sensitive internal capabilities.

Weapon layer

Tools are visible as governed capability, not public promises

Search, fetch, MCP, crawler, and source-risk entries can be described publicly without exposing live internal tools.

Registry first

governed

Tool capability, risk, cost, and status are described before a task can use them.

Review-gated use

safe default

High-risk tools stay candidate or disabled unless future gates explicitly allow them.

Boundary

MCP information on public pages remains context, not a public tools/list claim.

Evidence chain

Reports must explain what the evidence can and cannot support

Tool output is normalized into intelligence items, evidence records, limitations, and audit material before a report can be reviewed.

Source context

Each source can carry type, status, quality, relevance, and limitation fields.

Evidence records

Intelligence items become reportable evidence rather than disconnected notes.

Audit trail

Internal reviewers can inspect the reason a claim was allowed, limited, or rejected.

Report workflow

Drafts are built for review before any future delivery path

The platform can compose internal report drafts from evidence, but the current alpha does not send or sell customer reports.

Customer-readable draft

not delivered

Shorter report format for operators to inspect as a possible future customer artifact.

Audit report

internal

Full evidence, failure, conflict, cost, and review status stays available internally.

Boundary

Current report pages are samples for product explanation. They are not approved customer deliverables.

Admin cockpit

Internal operators get the operational view

The public site describes the operator cockpit without linking visitors into protected admin routes.

Reports

internal

Internal report drafts and audit material remain owner-only.

Evidence

internal

Evidence records, quality, and visibility stay behind the operator boundary.

Human Review

manual

Manual review controls are not exposed on the public site.

Finance

draft

Cost notes remain draft-only with no real billing.

Status

owner-only

Runtime status is monitored internally after production updates.

Security

guarded

Public/internal boundary checks remain part of the operator workflow.

Safety boundary

Public trust depends on visible limits

The production public site keeps strong capability private and presents only the safe, reviewable product surface.