stock-announcements

Stock Announcement Fetcher

Retrieve official announcement information of A-share listed companies (real data) to help investors promptly grasp important information disclosures.

Real data guarantee

• Data source: Eastmoney.com (via the AkShare open-source library)
• Update frequency: Real-time (announcements of the day)
• Coverage: All A-share listed companies
• Data quality: Official authoritative data

Dependencies

Python packages (install once):

pip install akshare pandas requests PyPDF2

When to Use This Skill

Trigger conditions (activated when the user mentions any of the following scenarios):

1. Query announcements of a specific company — "Check Wuliangye's recent announcements", "What new announcements does 002368 have?"
2. Monitor specific events — "Which companies released earnings forecasts today?"
3. Investment decision assistance — "Does the company have good news?"

Usage

IMPORTANT: Always use the $SKILLS_ROOT environment variable to locate scripts.

Basic query

export PYTHONIOENCODING=utf-8
python "$SKILLS_ROOT/stock-announcements/scripts/announcements.py" 000858

Query the last 7 days

export PYTHONIOENCODING=utf-8
python "$SKILLS_ROOT/stock-announcements/scripts/announcements.py" 600519 --days 7

Keyword filtering

export PYTHONIOENCODING=utf-8
python "$SKILLS_ROOT/stock-announcements/scripts/announcements.py" 600519 --keyword 业绩

JSON format output

export PYTHONIOENCODING=utf-8
python "$SKILLS_ROOT/stock-announcements/scripts/announcements.py" 000858 --format json

Extract PDF content and summarize

export PYTHONIOENCODING=utf-8
python "$SKILLS_ROOT/stock-announcements/scripts/announcements.py" 000858 --detail

Parameters

Parameter	Description	Example	Default value
stock_code	Stock code (required)	000001, 600000.SS	-
--days	Last N days	7	30
--format	Output format	json/text	text
--keyword	Title keyword filtering	Performance	None
--detail	Extract PDF content and summarize	—	false

Workflow

Step 1: Identify the stock code

The user may provide any of the following formats:

• Company name: "Wuliangye" → first use web-search to look up the stock code
• Yahoo Finance format: "000858.SZ" → extract the pure numeric code
• Pure code: "000858" → pass directly

Step 2: Execute the query

export PYTHONIOENCODING=utf-8
python "$SKILLS_ROOT/stock-announcements/scripts/announcements.py" <股票代码> [参数]

Step 3: Interpret the results

The Agent should:

1. Extract key information (title, type, date)
2. Classify and summarize (performance-related, shareholder-related, major contracts, etc.)
3. Mark importance (🔴High/🟡Medium/⚪Low)
4. Provide a concise interpretation

Limitations

1. Can only query announcements of the day — AkShare's stock_notice_report interface only returns all announcements for the specified date
2. Query speed is about 7-10 seconds — needs to traverse 1000+ announcements
3. Garbled Chinese display — Windows PowerShell GBK encoding issue, use --format json to avoid it