HKEX Company Announcements Scraper

Collect official HKEX company announcements as structured filing metadata for investment research, compliance review, and recurring regulatory monitoring. Choose a Hong Kong publication-date window and receive announcement IDs, issuer identifiers, timestamps, titles, categories, and direct official document links.

This Actor reads public HKEXnews JSON. It does not download PDFs, extract filing text, summarize documents, retrieve paid quotes, or make investment recommendations.

Why use this Actor?

Turn a disclosure search into consistent rows that can be joined to a watchlist or stored in a research database. Stable announcement IDs support downstream deduplication; zero-padded stock-code arrays preserve identifiers; publication timestamps carry the Hong Kong UTC+08:00 offset. Filtering and limits apply before result billing.

No HKEX account, proxy, paid data subscription, or external API key is needed for the supported public endpoint. The Actor has no automatic paid proxy fallback.

Who is it for?

• Analysts compiling daily filing watchlists and links for source review.
• Compliance teams retaining metadata trails of public company disclosures.
• Data engineers loading regulatory headlines into spreadsheets and warehouses.
• Automation builders scheduling recurring collection with their own alert rules.

The Actor returns a snapshot. Scheduling, cross-run change detection, and notifications belong to your downstream workflow, not an internal alerting service.

Get started

1. Set fromDate to a publication date in YYYY-MM-DD format.
2. Optionally set toDate, stock codes, or a title keyword.
3. Choose a result cap and run the Actor.
4. Inspect the Announcements dataset tab and the collection summary.
5. Export JSON, CSV, Excel, or another format supported by Apify datasets.

Try a known populated date:

{
  "fromDate": "2026-10-23",
  "toDate": "2026-10-23",
  "language": "en",
  "maxItems": 20
}

Input reference

Input	Default	Behavior
fromDate	Today in Hong Kong	Inclusive first publication date; no future dates.
toDate	fromDate	Inclusive last publication date. Maximum window: 31 days.
stockCodes	All	Up to 100 numeric strings. 700 becomes 00700. ANY supplied code may match ANY code on a filing.
keyword	No filter	Case-insensitive contiguous substring in the decoded title only. Input-edge whitespace is trimmed; internal spaces are literal.
language	en	Official English (en) or Traditional Chinese (zh) section; not machine translation.
maxItems	100	Global accepted unique row cap, 1–10000. No zero/unlimited mode.

Stock codes and keyword combine with AND. Both filters run locally against official results, so highly selective searches can take longer than unfiltered ones. A stock code is not an HKEX internal stock ID.

Empty input searches today in Hong Kong. This can legitimately return no announcements on quiet dates. For recurring monitoring, explicitly supply your desired date window or update the input in your scheduler. The Actor does not infer yesterday or persist an incremental cursor.

HKEX company announcements output

One dataset row represents one official announcement ID in the selected language section. Rows are deduplicated within a run. Later dates are processed first, with official newest-first ordering within a day.

Field	Meaning
announcementId	Official HKEXnews ID, kept as a string.
stockCodes	Array of zero-padded stock codes; empty if the source supplies no issuer code.
stockName	Official stock short name, or empty when absent.
publishedAt	ISO timestamp with Hong Kong +08:00 offset.
title	Official title with HTML entities decoded.
category	Official category and headline classification text.
documentUrl	Direct official document URL. The file is not downloaded.
fileType	Source format label, such as PDF or HTM.
fileSize	Source file-size label, not a measured download size.
language	Requested en or zh section.
scrapedAt	UTC collection timestamp.

Actual source record from the 2026-10-23 window (collection time varies):

{
  "announcementId": "12382302",
  "stockCodes": ["01236"],
  "stockName": "LDROBOT",
  "publishedAt": "2026-10-23T22:59:00+08:00",
  "title": "H Share Award Scheme",
  "category": "Documents on Display",
  "documentUrl": "https://www1.hkexnews.hk/listedco/listconews/sehk/2026/1023/2026102301723.pdf",
  "fileType": "PDF",
  "fileSize": "585KB",
  "language": "en",
  "scrapedAt": "2026-10-25T07:15:00.000Z"
}

The SUMMARY key-value-store record reports requests, retries, response bytes, unique scanned rows, delivered saved rows, dates visited, and limitReached for the requested result cap. billingLimitReached identifies an SDK billing stop; completionReason is billing_limit, max_items, or source_exhausted. A billing-limited result is partial, not full date-window coverage. Platform termination or timeout may prevent a final summary from being written.

How much does it cost to collect HKEX announcements?

Pay-per-event pricing has a $0.015 start fee and an item event for each delivered announcement row. BRONZE pricing is $0.0457 per announcement. Per-announcement rates are $0.052555 on FREE, $0.035646 on SILVER, and $0.02742 on GOLD, PLATINUM, and DIAMOND. No separate charge applies to document links, filters, scanned-but-rejected rows, duplicates, or retries. A zero-result run still incurs the start fee after valid input is accepted.

At the listed BRONZE rate, estimated Actor-event totals are $0.0607 for one result, $0.472 for 10 results, and $4.585 for 100 results, including the start fee. Tier eligibility depends on aggregate monthly Apify Store spend, not this Actor's individual result count. Apify invoices and any separately applicable platform charges remain authoritative.

Coverage and limits

• The route queries the official SEHK listed-company title-search corpus. No separate GEM-only selector or claim of a complete cross-board archive is provided.
• Date inputs are accepted from 1999-04-01 to today, but historical availability depends on the live index. No full-archive coverage guarantee is made.
• Date windows are at most 31 days; split longer research periods into separate runs.
• Each day has a 10000-row upstream scan safety cap. If unseen rows remain, the run fails with a clear message rather than silently claiming completeness.
• A capped run is a bounded subset, not the entire date window. Billing limits and timeout can also stop collection.
• English and Chinese sections can differ; this Actor does not pair translations.
• It returns all indexed filing categories available on this route, including records without an issuer code. A stock-code filter excludes such records.

Reliability and troubleshooting

Transient network, 429, and server errors receive up to three total attempts per request, with a 30-second request timeout and bounded backoff. Permanent HTTP errors and changed JSON shapes fail rather than masquerading as an empty feed. No browser or proxy recovery is enabled.

Why are there no results? Check the date, language, exact stock codes, and literal title substring. Compare with the official HKEXnews search. A weekend or selective filter may naturally be empty.

Why did the run fail after producing rows? An upstream error, timeout, malformed record, or safety cap can occur after partial output. Treat rows from a failed run as partial, investigate the log, and narrow the window. Do not treat a nonempty dataset as proof of a complete successful export.

Does a keyword search document text? No. Only the decoded title is tested; category and PDF text are not searched.

Integrations

Use Apify schedules for recurring collection, then an integration or webhook to send successful-run rows to Google Sheets, Make, n8n, or your own database. Use announcementId plus language as a downstream deduplication key across overlapping daily pulls. Separate document-review systems can follow the links; this Actor does not read the files.

For compliance trails, retain the timestamp, ID, category, and official URL alongside your own ingestion date. Build alert rules downstream, and keep human review for legally or financially significant decisions.

API usage

Start one run using your Apify token:

curl -X POST \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  'https://api.apify.com/v2/acts/automation-lab~hkex-company-announcements-scraper/runs' \
  -d '{"fromDate":"2026-10-23","maxItems":100}'

Keep the returned run ID. Check that same run to terminal success, then fetch its defaultDatasetId with bounded dataset pagination. Starting a run returns run metadata, not the announcement rows.

JavaScript with apify-client:

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('automation-lab/hkex-company-announcements-scraper')
  .call({ fromDate: '2026-10-23', maxItems: 100 });
if (run.status !== 'SUCCEEDED') throw new Error(`Run ${run.id}: ${run.status}`);
const { items } = await client.dataset(run.defaultDatasetId).listItems({ limit: 100 });
console.log(items);

Python with apify-client:

import os
from apify_client import ApifyClient
client = ApifyClient(os.environ['APIFY_TOKEN'])
run = client.actor('automation-lab/hkex-company-announcements-scraper').call(
    run_input={'fromDate': '2026-10-23', 'maxItems': 100})
if run['status'] != 'SUCCEEDED':
    raise RuntimeError(run['status'])
print(client.dataset(run['defaultDatasetId']).list_items(limit=100).items)

MCP setup

Use the Actor-specific hosted endpoint; private Actors may not be discoverable until publication and appropriate authorization.

Claude Code

claude mcp add --transport http apify \
  "https://mcp.apify.com?tools=automation-lab/hkex-company-announcements-scraper"

Claude Desktop, Cursor, and VS Code

Use the equivalent HTTP server configuration supported by your client's version, with Apify authorization configured securely:

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com?tools=automation-lab/hkex-company-announcements-scraper"
    }
  }
}

Example prompt: “Collect 20 English HKEX announcements for 2026-10-23 and show titles, stock codes, publication times, and official document links.”

Example prompt: “Collect announcements for stock code 00700 on 2026-10-23, then return only rows whose title contains results.”

Discover actual tool names and argument schemas with tools/list. Actor selection also exposes run/data helpers, not exactly one tool. For an authorized invocation, start once and retain the run ID. Follow that same run with bounded backoff (2, 4, 8 seconds, capped at 10) and a 120-second consumer deadline; configure client timeout above the chosen server wait plus transport margin. If the deadline passes, report pending with the run ID rather than restarting.

After success, read untransformed dataset pages of 20 rows with explicit offsets and fields. A suggested model-context budget is 100 source rows or 64 KiB of serialized UTF-8 content, whichever comes first. Enforce byte limits host-side before model injection; if the client cannot intercept oversized responses, do not claim a hard byte guarantee. Report continuation offsets and partial/truncated status honestly. Full exports remain in storage outside model context. No numeric token-saving claim is made.

Legality and data handling

This is an independent automation-lab tool, not affiliated with or endorsed by HKEX. Source names identify the public corpus only; the icon is an original generated document/data illustration, not an official exchange logo. Use Apify's standard terms and comply with source rights and applicable law. Public filings can contain regulated information; verify the official document before relying on a headline.

The runtime uses no AI service and sends no announcement data to a model provider. Apify hosts inputs, logs, and per-run datasets under your account's retention settings; you control deletion through Apify storage/run controls. There is no persistent cross-run cache or copied document archive. Do not place credentials or private information in title filters.

Failed operations send sanitized diagnostic input, exceptions, and Actor/build/run IDs to our private GlitchTip service for repair. Secret fields and URL queries are removed; reports are retained for 30 days. General logs contain counters and safe date/status information, not document bodies or request credentials.

FAQ

Can I obtain PDF text or AI summaries? No. This product exports metadata and filing URLs only.

Can I get every filing in a month? Use windows of at most 31 days and a sufficient result cap; confirm successful completion and source coverage. Split busy windows into smaller runs when needed.

Can I use it as a real-time alert service? You can schedule recurring snapshots and compare IDs downstream. No real-time delivery guarantee, webhook alert generation, or internal incremental state is included.

Support

For technical help, use the Actor's Apify Issues tab with a safe input sample, run link, expected result, and observed behavior. Never post tokens or private data.

Related workflows

This Actor is intentionally standalone for HKEX metadata. Separate jurisdictions require their own source-specific collectors and validation; it does not combine regional feeds or imply equivalent coverage. Connect this dataset to your existing spreadsheet, warehouse, or filing-review pipeline.