Introducing a new way to review your PRs.

Read the announcement

koala73 / worldmonitor

Real-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface

View on GitHub

29 documents · Updated Nov 20, 2026

docs / travel-risk-assessment

Travel-risk assessment workflows

Travel-risk assessment turns a selected itinerary into a sourced draft for analyst review. It accepts countries in travel order with optional cities, dates, and a task label, reads current monitoring signals, and delivers the same assessment through the web export and the MCP tool. It is a child product of Intelligence products and relates to Intelligence briefs and analysis, Company monitoring intelligence, Forecasting, scenarios, and resilience, Domain API routes, and Access control and subscriptions.

Product flow

The workflow combines a focused dashboard scope with extra trip detail, reads the signals that cover each destination, and produces one reviewable document. The document keeps source dates and source status visible so missing or old data does not look like a clean risk result.

flowchart LR
    A[Focused countries and date window] --> B[Trip details]
    B --> C[Validated itinerary]
    C --> D[Government and monitoring signals]
    D --> E[Ratings and country findings]
    E --> F[Markdown and structured assessment]
    F --> G[Web export]
    F --> H[MCP response]

Access and destination inputs

The browser starts an assessment from the countries and date window in the current focus. The dialog lets the user add or remove destinations, keep travel order, add up to five cities per country, set arrival and departure dates, and add a task label. The input describes the trip only. It does not collect traveller names, addresses, or contact details.

The dialog checks simple date and count errors before sending the request. The server applies the same contract again and also checks country codes, duplicate countries, travel order, and the trip window. A focus with more than six countries reports the omitted countries instead of silently adding them to the assessment.

References: src/…/TripAssessmentDialog.ts• L18-27 src/…/travel-risk-export.ts• L15-45 src/…/travel-risk-export.ts• L68-112 shared/report-handoff.ts• L1-9 api/…/travel-risk-tool.ts• L28-31

Signal aggregation

The assessment reads the feeds that can support the itinerary at request time. It combines government advice with security events, hotspot and instability views, natural hazards, health reports, airport status, and internet outage signals. It also uses recorded seasonal hazard history when the trip has usable dates.

Each source read has a time limit and a freshness check. The result records whether a source is current, out of date, unavailable, or not used. Country and city matching gives the analyst context; it does not turn the document into a precise route or venue safety guarantee.

References: api/…/_travel-risk-document.ts• L29-66 api/…/_travel-risk-document.ts• L83-101 api/…/travel-risk-tool.ts• L281-312 api/…/_travel-risk-document.ts• L1090-1097

Assessment rendering

The product renders a sourced draft rather than a final travel decision. It gives each country a rating based on the higher of the US State Department and Australian DFAT levels, then reports other advice scales beside that rating. The other sections provide evidence and context but do not change the rating.

The document has a summary followed by government advice, security, natural hazards, health, getting in and out, and communications. It adds empty analyst sections for topics that the feeds do not cover. Each finding names its source and source date, while unavailable data is marked as unavailable rather than treated as no risk.

References: api/…/_travel-risk-document.ts• L135-169 api/…/_travel-risk-document.ts• L458-485 api/…/_travel-risk-document.ts• L1103-1151 api/…/_travel-risk-document.ts• L1196-1245

Delivery and export

The MCP tool is the shared product path for agent requests and structured responses. The web API calls the same tool executor, so the browser export and MCP output use the same source reads and assessment content. The API returns a no-store Markdown or HTML attachment.

The browser offers Markdown download, HTML suitable for Word, and browser printing for PDF. HTML is derived from the same Markdown draft. The system keeps the draft in the active request and does not store the trip or the generated file.

References: api/…/travel-risk-tool.ts• L450-504 server/…/travel-risk-export.ts• L1-19 server/…/travel-risk-export.ts• L172-202 server/…/travel-risk-html.ts• L1-21 src/…/TripAssessmentDialog.ts• L366-415

Entitlement enforcement

Travel-risk assessment is a plan feature, not a general tier capability. The supported grant is named travelRiskAssessment and is included in Pro Business, API Business, and Enterprise plans.

Browser presentation gate

The focus-bar action is shown only when the merged account entitlement has an active tier, an explicit travel-risk grant, and a valid coverage time. This gate improves the user experience, but it does not protect the product because browser presentation can be bypassed.

References: shared/travel-risk-access.ts• L1-14 shared/travel-risk-access.ts• L16-30 src/…/entitlements.ts• L351-358 src/…/focus-ui.ts• L25-64

Server authorization gate

The web export requires a signed-in browser session and checks the entitlement before it reads the request body or source feeds. The MCP path resolves the caller, rechecks the current entitlement, and passes the feature result into dispatch. A missing flag or expired grant is denied. An entitlement service outage is returned as a retryable service error instead of an upgrade message.

The web export has its own account request budget because one assessment reads many cached sources. MCP calls use the caller's MCP access and quota rules. Both surfaces therefore enforce access at the server boundary even when the browser action is hidden.

References: server/…/travel-risk-export.ts• L97-155 api/travel-risk-assessment.ts• L20-53 api/…/auth.ts• L733-864 api/…/handler.ts• L1366-1388 api/…/dispatch.ts• L671-679

Review boundaries

The assessment is a starting point for duty-of-care or trip review work. It does not claim to cover every local risk, and its rating is not a recommendation to travel. Analysts must complete the marked gaps and inspect the source dates before using the document for a decision.

The source method also states important limits: security counts use a recent fixed window, some events use country bounds, listed cities only mark matching signals, and seasonal history is not a forecast. These limits belong with the assessment because they define how its evidence should be read.

References: api/…/_travel-risk-document.ts• L59-66 api/…/_travel-risk-document.ts• L1042-1097 api/…/_travel-risk-document.ts• L1196-1245

Change checklist

When these files or sections are changed, remember to consider these:

  • Keep the browser draft shape, MCP input schema, and server parser aligned for country order, dates, city limits, and task labels.
  • Keep the MCP executor and web export on the same document builder so Markdown, HTML, and structured output cannot diverge.
  • Keep source status and freshness semantics aligned between country sections, the summary, and the sources table.
  • Keep the merged entitlement field and its fail-closed behavior aligned across the browser gate, MCP pre-check, dispatch guard, and export route.
  • Keep the six-country handoff limit aligned with the focus handoff so omitted countries remain visible to the user and agent.

Important failure modes

TriggerConsequenceGuard
Entitlement flag is absent, false, or expiredBrowser action is hidden and protected delivery returns an upgrade denialShared entitlement check requires an explicit active grant
Entitlement service cannot verify the accountA valid customer could be misclassified as ineligibleServer returns a retryable unavailable response and does not read sources
All travel-risk feeds fail or are unavailableNo assessment is producedSource aggregation requires at least one usable input
A single feed is stale or unavailableIts section can be incomplete or misleading if treated as currentThe document marks source status and says unavailable is not no risk
Trip input breaks schema or size limitsNo source read occurs and the caller receives validation feedbackClient and server validate the itinerary before building
More than six focus countries are selectedSome destinations would be left out of the assessmentThe handoff records omitted countries and the dialog shows them

Key files

FileWhy to read it
src/…/TripAssessmentDialog.tsShows the user-facing trip form and the three export actions.
src/…/travel-risk-export.tsDefines the browser trip contract and maps server responses to user outcomes.
api/…/travel-risk-tool.tsDefines the MCP product contract, source gathering, and execution boundary.
api/…/_travel-risk-document.tsExplains the assessment shape, ratings, source status, and rendered sections.
server/…/travel-risk-export.tsShows the protected web export flow and its shared execution path.
server/…/travel-risk-html.tsShows how the shared Markdown draft becomes Word- and print-friendly HTML.
shared/travel-risk-access.tsDefines the common active-entitlement rule used by presentation and server checks.
shared/report-handoff.tsDefines how a focused dashboard scope becomes an assessment itinerary.
api/…/auth.tsExplains MCP identity, entitlement, billing verification, and quota decisions for gated tools.