CBI Data MCP
A read-only, evidence-qualified research service over eleven citizenship-by-investment programmes and 44 short-stay passport travel-rule slots — every response carries release pinning, coverage counts, and warnings that missing context never establishes eligibility or a travel waiver.
list_programmes — browse programmes filtered by jurisdiction, stage (published route vs published with limits) and kind, with cursor pagination.
get_programme — full programme detail: options, required fields with gap/partial/supported states, and per-claim evidence, verification and review status.
compare_programmes — side-by-side comparison of 2–4 programmes, optionally scoped to a scenario, including cost comparisons.
get_programme_costs — exact known subtotals by currency, an explicitly null comprehensive total, missing and excluded items, and recurring obligations.
get_travel_rule — passport-to-destination short-stay rules that account for transport, purpose, document type and qualifying visa, or report
not_covered.get_source — source metadata: publisher, authority type, rights (link-only), retrieval health and hash, review cadence, and the claims it supports.
get_changes — paginated diffs between two dataset releases, filterable by programme, with revision IDs.
get_coverage — dataset-wide counts (claims, sources, travel slots, gaps, disputed claims), scope, review-due count and monitoring state.
Shared guarantees — pin any release, optionally request evidence blocks, and get read-only, idempotent results requiring no account, storage or API key.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@CBI Data MCP compare Malta and Saint Lucia citizenship-by-investment programmes"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
CBI Data
Current software 0.2.1, dataset 2026-11-15.3: dated passport issuance and programme activity. Nine read-only tools; 34 activity subjects and 72 observations, including explicit unknowns and attributed estimates. The retail legal dataset remains eleven programmes.
Dated, evidence-qualified investment citizenship research for eleven programmes. The original public compilation is authoritative for its migrated CBI.me claims. It includes 44 ordinary-passport policy lookup slots for the UK, US, Canada and Schengen short stays. These are qualified slots, not 44 unconditional waivers. Every required programme field has evidence or an explicit incomplete state.
The compilation retains dated editorial reviews with partial verification; it does not claim a newly completed legal audit. Conflicting Saint Lucia prices, São Tomé adult-dependant issuance limits, US issuance exceptions and conditional Canadian air eTA access remain explicit. Costs return exact known subtotals by currency and a null comprehensive total. Eligibility and admission are separate.
The latest release adds Caribbean transition announcements: St Kitts remains open with limits during its negotiated wind-down; other country positions retain their stated scope.
Reproduce a pinned release
Requires Node 24 LTS. No maintainer account, private repository or API key needed.
git clone --branch data-2026-11-14.2 https://github.com/heathermhuang/cbi-data.git
cd cbi-data
npm ci --ignore-scripts
npm run check
node examples/compare.mjs
node examples/travel-rule.mjs
node examples/mcp-client.mjsUse the GitHub Release dataset archive and SHA256SUMS for static JSON/CSV.
The package tarball can be installed locally; npm registry publication is not
required. Never substitute an unversioned installer for the pinned example.
curl -fsS 'https://cbi.me/api/v1/travel-rules?passport=ATG&destination=CAN&release=2026-10-25.2'
curl -fsS 'https://cbi.me/api/v1/coverage?release=2026-10-25.2'Remote MCP: https://cbi.me/mcp (Streamable HTTP, anonymous public reads).
Local MCP: node /absolute/path/to/cbi-data/src/cli/stdio.mjs after the pinned
checkout and build above. Stdout contains protocol messages only.
Eight tools: list_programmes, get_programme, compare_programmes,
get_programme_costs, get_travel_rule, get_source, get_changes, get_coverage.
The hosted service also retains its four existing guide tools. The standalone
adapter has no private workspace tools, accounts, bindings or storage.
See API, MCP, methodology, coverage, maintenance, releases and contributing.
Original code MIT; original compilation CC BY 4.0. Linked sources retain rights.
Available Tools
9 toolscompare_programmesBRead-onlyIdempotent
Read-only compare programmes. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: ids (required), include, release, scenario. Returns: data, coverage, evidence, warnings, +4 more
get_changesCRead-onlyIdempotent
Read-only get changes. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: to (required), from (required), limit, cursor, +3 more. Returns: data, coverage, evidence, warnings, +4 more
get_coverageCRead-onlyIdempotent
Read-only get coverage. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: release. Returns: data, coverage, evidence, warnings, +4 more
get_programmeBRead-onlyIdempotent
Read-only get programme. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: id (required), include, release. Returns: data, coverage, evidence, warnings, +4 more
get_programme_costsCRead-onlyIdempotent
Read-only get programme costs. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: id (required), option, include, release, +1 more. Returns: data, coverage, evidence, warnings, +4 more
get_programme_statisticsCRead-onlyIdempotent
Read-only get programme statistics. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: id (required), limit, cursor, include, +2 more. Returns: data, coverage, evidence, warnings, +4 more
get_sourceCRead-onlyIdempotent
Read-only get source. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: id (required), include, release. Returns: data, coverage, evidence, warnings, +4 more
get_travel_ruleCRead-onlyIdempotent
Read-only get travel rule. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: passport (required), destination (required), include, purpose, +4 more. Returns: data, coverage, evidence, warnings, +4 more
list_programmesCRead-onlyIdempotent
Read-only list programmes. Preserve evidence dates, conditions, coverage gaps and warnings. Missing context never establishes eligibility or a travel waiver.
Inputs: kind, limit, stage, cursor, +3 more. Returns: data, coverage, evidence, warnings, +4 more
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 9 tools
Each tool targets a distinct facet of the CBI data domain (listing, detail, comparison, costs, travel rules, sources, changes, coverage, statistics), so an agent can mostly pick correctly from names alone. However, get_programme vs get_programme_costs vs get_programme_statistics have overlapping scope, and the identical boilerplate descriptions provide no differentiating guidance beyond the name.
All nine tools follow a consistent snake_case verb_noun pattern (list_programmes, get_travel_rule, compare_programmes), with a clear get_/list_/compare_ prefix scheme. No mixing of conventions or vague verbs.
Nine tools is a well-scoped set for a read-only reference data server, with each tool mapping to a distinct data facet. Nothing feels padded or missing at the surface level.
The read-only surface covers the lifecycle of data lookup well: listing, detail, comparison, costs, travel rules, provenance (get_source), change history, coverage, and statistics. The main gap is the absence of any search/filter tool or batch lookup, though agents can work around this via list_programmes.
Maintenance
Related MCP Connectors
Mira, by Mirabello: verified, source-cited data on citizenship and residency by investment, golden visas, immigration pathways, visa requirements, naturalisation and tax residency across 190+ countries. 68 tools, remote, no key needed. Built and operated by Mirabello Consultancy (Zurich, Dubai, Hong Kong): https://www.mirabelloconsultancy.com/mcp. General information, not legal, tax or immigration advice.
CitedTravel gives AI agents and developers travel facts they can cite. It covers plug types and voltage, currency, emergency numbers, driving side, public holidays (3 years) and tipping norms for 195 countries, plus a visa/entry pilot for GB, US and JP passport holders. Every value comes with its sources (publisher, URL, license, verbatim quote), a last-verified date and a 0-1 confidence score with reasons. Data comes only from sources that allow commercial reuse, such as Wikidata, Unicode CLDR, UK FCDO, the US State Department and Japan MOFA. Entry answers quote official government pages and are decision support, never a go/no-go verdict. Remote MCP server (Streamable HTTP, 9 read-only tools) and REST API (OpenAPI 3.1). Answers in English, Japanese, Simplified and Traditional Chinese. Free: static facts need no key; a free key adds 1,000 calls/month. Paid: Builder $19/month, Pro $99/month, or x402 pay-per-call in USDC on Base.
Pay-per-call government data over x402: provenance chains, Ed25519 attestations, no API keys.
Relocation, residency, visa, company-formation and crypto-tax data for about 170 countries.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceDocument verification for AI agents: forensic authenticity signals for PDFs and images, field extraction, Australian identity checks, adverse-media and sanctions screening, AU/NZ government tender search, AI-text detection, and citation verification. Hosted with a free anonymous tier; the repo ships a Dockerfile that bridges to the live endpoint.1MIT
- AlicenseAqualityBmaintenanceExposes fourteen read-only tools over the HEY Research public API, letting assistants look up Robinhood Chain projects, builders, shipped work, relationships and market context, with each answer tagged FACT, DERIVED or UNKNOWN and unmeasured fields omitted rather than zeroed. Runs over stdio or hosted Streamable HTTP, and gives no trade instructions.131MIT
- AlicenseNot gradedqualityBmaintenanceExposes five read-only tools that let clients query a public knowledge base of cognitive models — browsing taxonomy and categories, searching models, retrieving a model's details, exploring its lineage and relationships, and inspecting cited sources with verification status. It runs as a standalone Node.js service over local stdio or self-hosted Streamable HTTP, so assistants can ground answers in real, cited model data.1Creative Commons Zero v1.0 Universal
- AlicenseAqualityBmaintenanceProvides AI agents with 12 real-time tools for search synthesis, grounded research dossiers, claim verification, citation generation, sentiment analysis, expert ranking, timeline creation, and competitive landscape mapping over Streamable HTTP or stdio.112MIT