The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Hk Companies listing page.
Hong Kong Companies Registry (香港公司註冊處) open data — a live feed of companies newly incorporated, registered, or renamed on the register, with a name/BR-number search and an HKEX ticker join.
Part of Pipeworx — an MCP gateway connecting AI agents to 1683+ live data sources.
hk_company_search({ name?, brn?, type?, since?, limit? }) — search by
company name (English or Chinese, substring) or exact BR number. Returns BR
number, English/Chinese name, type, incorporation/registration date, any
name-change date, and the matching HKEX ticker when the name resolves to a
listed stock.hk_company({ brn, since? }) — look up one company's full history (initial
incorporation/registration + any renames) by BR number.hk_new_companies({ since?, type?, limit? }) — companies newly incorporated
or registered since a date, most recent first.hk_company_name_changes({ since?, limit? }) — companies that changed their
registered name since a date, most recent first.Keyless. data.gov.hk and cr.gov.hk are public, unauthenticated open data.
The Companies Registry's open-data landing page (cr.gov.hk/en/open-data/) is
404 and data.gov.hk's own HTML dataset listing is client-rendered (zero
<a href>s in the page source), so the real resource URLs come only from
data.gov.hk's CKAN-style API, found via its org slug for the Registry:
That package lists ~350 weekly CSV resource pairs, one per week since 2024-12-30:
https://www.cr.gov.hk/docs/wrpt/RNC063/RNC063L_YYYYMMDD.csv — Hong Kong
local companies (newly incorporated / re-domiciled / renamed)https://www.cr.gov.hk/docs/wrpt/RNC063/RNC063F_YYYYMMDD.csv — non-Hong
Kong companies (newly registered / re-domiciled / renamed)YYYYMMDD is the Monday of the covered week; each file covers that Monday
through the following Sunday. Publication runs roughly a week behind (the
24-30 Aug 2026 week was published 2 Sep).
This pack fetches the live resource list from package_show on every call
(no guessed URL pattern), then only fetches the CSV weeks that can contain a
match, in parallel, bounded to 20 weeks per company type per call so one
request can't fan out into hundreds of upstream fetches. hk_company_search
and hk_company default to a 12-week lookback unless since is given (as
far back as 2024-12-30, the dataset start); hk_new_companies and
hk_company_name_changes default to the 7 days before the latest published
week.
Local (RNC063L): Seq, Current Company Name in English, Current Company Name in Chinese, BR Number, Date of Incorporation / Re-domiciliation Date, Date of Change of name — English and Chinese are already separate columns.
Non-HK (RNC063F): Seq, Current Corporate Name / Other Corporate Name, Current Approved Name for Carrying on Business in H.K., BR Number, Date of Registration, Date of Change of name — there is no separate Chinese-name
column. A non-HK company with both an English and a Chinese name gets two
physical rows sharing the same BR number and Seq, one name per row in the
single "Current Corporate Name / Other Corporate Name" column. This pack
merges those pairs by BR number (routing CJK text to name_zh, everything
else to name_en) rather than surfacing them as two different companies.
This is the only company-level open dataset the Companies Registry
publishes. Its other three data.gov.hk packages under org hk-cr
(crdata-stat-non-hk-companies, crdata-stat-local-companies-incorporated,
crdata-stat-local-companies) are aggregate monthly counts (public /
private / guarantee company totals), not per-company records — checked live,
they carry no name, BR number, or address fields. data.gov.hk's org for the
Inland Revenue Department (hk-ird) publishes zero datasets, so Business
Registration data has no open API either.
There is no open, keyless API that returns a company's registered address or status for an arbitrary Hong Kong company. That is the Companies Registry's paid Cyber Search Centre (ICRIS) product — a per-document commercial search, not open data. So:
<any company>"
for a company outside that window (most established companies, including
every HKEX-listed blue chip) — there is genuinely no open source for that.
Every tool description says so rather than let a caller assume more than
the data supports.data.gov.hk (checked: zero
results searching "money lender", "TCSP", "trust company", "licensed money
lenders" against the CKAN API) — not built.hk_company_search and hk_company call the same keyless Yahoo Finance
search endpoint @pipeworx/hk-stocks's hk_resolve_symbol uses (called
directly here, not cross-pack, since packs don't call each other at
runtime), restricted to .HK-suffixed symbols, and only attach a ticker on a
normalized exact-name match (stripping "Limited"/"Ltd"/"Co"/"Holdings"/
"Group" and non-alphanumerics) — never a fuzzy guess. In practice this fires
rarely: a company appearing in this feed was newly incorporated, registered,
or renamed in the last ~20 months, and HKEX-listed companies are almost
always older than that.
The feed publishes roughly a week behind. A since date newer than the
latest published week returns count: 0 with a data_lag_note field
explaining that this is publication lag, not "no new companies" — check that
field before reading a zero as a clean result.
Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):
tools/list at https://gateway.pipeworx.io/hk-companies/mcp returns the tools in the table
above plus the shared Pipeworx meta-tools — ask_pipeworx,
discover_tools, search_within, remember/recall and the rest of the
gateway-wide set. So the tool count you see is larger than this table: a
single-pack endpoint currently lists roughly 30 shared tools alongside the
pack's own. The connection's initialize response states its exact scope, and
is the authoritative answer for a given day.
This is deliberate, not multiplexing by accident. The meta-tools are what let a
scoped connection answer a question this pack does not cover — via
ask_pipeworx, which routes across the whole catalog — without you adding a
second MCP server. There is currently no way to mount a pack endpoint without
them; if the extra schemas cost you more context than the routing is worth,
connect to the full gateway once rather than to several pack endpoints.
Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:
Both URLs reach the same gateway and the same 1683+ data sources. The
only difference is which pack's tools are listed directly; ask_pipeworx
reaches all of them from either one.
No account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/hk_company_search. Find one: POST https://gateway.pipeworx.io/v1/tools/search_packs with {"query":"..."}.
This package also runs as a local stdio MCP server — no Pipeworx account, no gateway round-trip:
Or run it directly to confirm it starts:
It speaks MCP over stdin/stdout and answers initialize/tools/list/tools/call
for only this pack's tools — none of the shared meta-tools the gateway
connection above adds. Same source, same tools, no ask_pipeworx routing.
Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:
The gateway picks the right tool and fills the arguments automatically.
MIT