fd-cn-report
OfficialOne-line summary: MCP server for Chinese financial reports — resolve companies, fetch filings/PDFs, extract sections, run AI extraction, manage indicators, search Elasticsearch, and access HK/exchange/CSRC/ministry data.
Resolve CN-A-share companies by ticker/name and list/fetch CNINFO filings by form, category, or special report type (
get_company,list_filings,get_filing,list_report_types,get_special_report).Parse report outlines and extract sections by exact title, regex, or ordinal from PDFs/URLs (
list_outline,extract_section,get_section).Run LLM structured extraction over section text (
ai_extract).Bulk-index records into Elasticsearch and search with BM25 + filters/highlights (
index_records,search_reports,delete_index).Pull structured financials via akshare and three major statements as text (
get_financials,get_financial_statements).Browse, resolve, batch-extract, and audit financial indicators using the rule engine (
list_indicators,get_indicator,extract_indicators,extract_indicators_by_position,audit_rule_gaps).Manage on-disk report cache (
list_cache,clear_cache).Access HK stock data: company lookup, filings, financials, sections (
get_hk_company,list_hk_filings,get_hk_financials,get_hk_section).Access exchange-specific disclosures and investor Q&A for SSE/SZSE/BSE (
get_sse_*,get_szse_*,get_bse_*).Access CSRC regulatory data: announcements, IPO/merger review status, enforcement actions (
list_csrc_filings,get_csrc_ipo_review,get_csrc_merger_review,list_csrc_enforcement).Access ministry statistics: list sources, fetch NBS/ministry stats (
list_ministries,get_nbs_stat,get_ministry_stat).Launch a web dashboard to browse/filter/search 21,698 industry rules across 31 申万 L1 industries (
open_industry_rules_dashboard).
Provides Elasticsearch integration for storing and searching extracted financial report data, including bulk indexing of records, BM25 search with highlight snippets, and index deletion.
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., "@fd-cn-report 提取贵州茅台2023年报中的毛利率"
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.
fd-cn-report
Wire (柏讯) product line · the open-data supply line of FindData — Chinese financial reports MCP
English | 中文
MCP server for Chinese financial reports — 31 申万 L1 industry AI rule system, outline extraction, AI structured extraction, Elasticsearch store + search, and interactive rules dashboard.
One-click install
fd-cn-report auto-registers as a datasource when installed alongside
fd-open-data-mcp. To install the entire finddata stack (hub + every
datasource + ontology DB) in one go:
pip install "fd-open-data-mcp[data]" fd-polygon fd-cn-report
fd-open-data-mcp migrate \
&& fd-open-data-mcp import-catalog \
&& fd-open-data-mcp consume-concepts \
&& fd-open-data-mcp propose-bindings \
&& fd-open-data-mcp seed-entities \
&& fd-open-data-mcp generate-schedules \
&& fd-open-data-mcp register-discovered
fd-open-data-mcp serveTo run fd-cn-report's own MCP server standalone (without the hub):
uv sync && uv run python server.py — see Setup below.
Related MCP server: sfc-data-mcp
Industry Rules System
21,698 LLM rules covering 31 申万 L1 industries with per-section extraction instructions generated from real annual report PDFs. Each rule maps an indicator to its exact section position in the periodic report.
Coverage
章节 | 内容 | 说明 |
资产负债表 | 全部科目 (112+ rules) | 从合并资产负债表提取 |
利润表 | 全部科目 (130+ rules) | 从合并利润表提取 |
现金流量表 | 全部科目 (42+ rules) | 从合并现金流量表提取 |
管理层讨论 | 主营业务分析、成本、产销量、研发 | 从实际报告第三节提取 |
主要财务指标 | ROE、EPS、毛利率、分季度数据 | 从第二节提取 |
报表附注/重要事项 | 关联交易、担保、诉讼、资产减值 | 从第六节+第十节附注提取 |
股东信息 | 前十大股东、持股变动、分红 | 从第七节提取 |
公司治理 | 董事会、高管薪酬 | 从第四节提取 |
环境 | 排污、碳排放、能源消耗 | 从第五节提取 |
Industry-Specific Rules
6 industries have industry-specific rules on top of the universal set:
行业 | 特有指标 | 来源 |
银行 | 不良率、净息差、资本充足率、贷款五级分类 | 工商银行年报 |
房地产 | 合同负债、存货-开发成本、土储、销售面积 | 保利发展年报 |
电力设备 | 在建工程、产能利用率、应收账款 | 宁德时代年报 |
医药生物 | 研发费用、销售费用、在研管线、无形资产 | 恒瑞医药年报 |
非银金融 | 保费收入、赔付支出、偿付能力、新业务价值 | 中国平安年报 |
农林牧渔 | 存货-消耗性生物资产 | 隆平高科年报 |
Rule Generation Pipeline
Rules are generated from real annual report PDFs via parallel LLM calls:
# Generate rules for all 31 industries from real reports
python scripts/generate_rules_from_real_reports.py --max-concurrent 5
# Single industry
python scripts/generate_rules_from_real_reports.py --industry 801120
# Run industry-specific sections only
python scripts/generate_rules_from_real_reports.py --industry 801780 --llm-onlyDashboard
Interactive web UI to browse, filter, and search all 21,698 rules:
# CLI
python scripts/industry_rules_dashboard.py
# or
fd-cn-report
# MCP tool
open_industry_rules_dashboard(port=8888)Features:
Industry filter — dropdown with all 31 申万 L1 industries
Module filter — filter by balance_sheet, income_statement, etc.
Keyword search — indicator name, section, instruction text
Sortable columns — click header to sort
Pagination — 50 rules per page
Color-coded tags — module type visualization
Real-time stats — filtered count, industry count, module count
Tools (44 MCP tools)
Layer | Tool | Description |
Company API |
| Resolve ticker/name → company entry |
| List CNINFO disclosures by form/category + year | |
| One announcement's metadata + PDF URL | |
| Income/balance/cashflow via akshare | |
| 三大报表 as text from the annual-report PDF | |
|
| |
| Browse CNINFO disclosure category catalog | |
| Special-type report (招股说明书, 收购报告书, …) | |
HK Stock |
| Resolve HK stock by ticker/name |
| List HKEX filings | |
| HK financial statements | |
| HK report section extraction | |
Official-website (SSE/SZSE/BSE) |
| Resolve SSE-listed company by 6-digit ticker |
| List SSE disclosures from sse.com.cn | |
| SSE annual-report section extraction | |
| 上证e互动 investor Q&A | |
| Resolve SZSE-listed company by 6-digit ticker | |
| List SZSE disclosures from szse.cn | |
| SZSE annual-report section extraction | |
| 互动易 investor Q&A | |
| Resolve BSE-listed company by 6-digit ticker | |
| List BSE disclosures (BSE-native, CNINFO fallback) | |
| BSE annual-report section extraction | |
Official-website (CSRC) |
| List CSRC regulatory announcements |
| CSRC IPO (首发) review status | |
| CSRC 并购重组 review status | |
| CSRC enforcement actions | |
Ministry statistics |
| List supported ministry stat sources |
| Ministry stats page -> HTML tables | |
| NBS macro statistic by indicator code | |
PDF / AI / ES |
| Parse 目录 from report URL or PDF path |
| Body text by exact title / regex / ordinal | |
| LLM-structured extraction over section text | |
| Bulk index records into ES | |
| BM25 + filter search with highlights | |
| Drop | |
Cache |
| List cached reports |
| Evict cached reports | |
Indicators |
| Browse indicator rule set |
| One indicator's value | |
| All applicable indicators in one pass | |
| CSV-driven extraction | |
| Audit missing rules across industries/tickers | |
Dashboard |
| Start the rules web dashboard |
extract_indicators_batchis a Python convenience function (not an MCP tool) — see Concurrency.
Typical Chain
# 1. Resolve company → 2. find latest annual → 3. pull MD&A → 4. LLM-extract revenue table
co = get_company("600519")
# {"stock_code": "600519", "name": "贵州茅台", "org_id": "gssh0600519", "exchange": "sse", ...}
filings = list_filings("600519", form="年度报告", year=2023, limit=3)
# [{"announcement_id": "1219730876", "pdf_url": "http://static.cninfo.com.cn/.../*.PDF", ...}]
sec = get_section("600519", year=2023, section="管理层讨论与分析")
# {"text": "<full MD&A body>", "pdf_url": "...", "outline_entry": {...}, ...}
records = ai_extract(
text=sec["text"],
schema={"type": "object", "properties": {
"segment": {"type": "string"},
"revenue_2023": {"type": "string"},
}, "required": ["segment", "revenue_2023"]},
)
# {"records": [{"segment": "茅台酒", "revenue_2023": "139,989,000,000"}, ...]}Special Report Types
CNINFO exposes dozens of disclosure categories beyond the four periodic reports (招股说明书, 增发, 业绩预告, 收购报告书, 股权激励, …). Browse the catalog, then list or retrieve by category:
catalog = list_report_types()
# {"groups": [{"name": "定期报告", "categories": [...]}, {"name": "融资", ...}, ...], "count": 26}
list_report_types(group="融资")
# {"group": "融资", "categories": [{name: "首发", code: "category_sf_szsh", ...}, ...], "count": 6}
filings = list_filings("600519", category="首发", limit=3)
sec = get_special_report("600519", category="首发", section="募集资金运用")三大报表 (Three Major Financial Statements)
get_financials returns akshare's structured numeric tables. get_financial_statements
pulls the three major statement sections as text from the PDF:
stmts = get_financial_statements("600519", year=2023)
# {
# "stock_code": "600519", "company_name": "贵州茅台", "year": 2023,
# "form": "年度报告", "pdf_url": "...", "cached": False,
# "statements": {
# "income_statement": {"title": "2、 合并利润表", "outline_entry": {...}, "char_count": 4521, "text": "..."},
# "balance_sheet": {"title": "1、 合并资产负债表", ...},
# "cashflow": {"title": "3、 合并现金流量表", ...},
# },
# "missing": [],
# }Report Cache
Every report fetch goes through an on-disk cache under .cache/reports/.
First fetch downloads PDF + extracts text + outline; subsequent fetches read from disk.
list_cache()
# {"cache_dir": ".../.cache/reports", "count": 2, "entries": [...]}
clear_cache() # evict everything
clear_cache(stock_code="600519") # evict one company
clear_cache(stock_code="600519", year=2023) # evict one company + yearIndicators
The indicator engine profiles each company, filters applicable rules, and routes each indicator to akshare / report-section / computed / external.
# Preview → pull one → pull all → CSV
list_indicators(company="工商银行") # rules applicable to 工商银行
get_indicator("资本充足率", "工商银行", 2023) # one value
extract_indicators("工商银行", 2023) # all applicable, one PDF fetch
extract_indicators("工商银行", 2023,
indicators=["资本充足率","不良率"]) # subset
extract_indicators("工商银行", 2023, extractor_mode="python") # LLM-free
extract_indicators_by_position("工商银行", 2023) # CSV-drivenMulti-form support (年度报告 / 半年度报告 / 第一季度报告 / 第三季度报告):
extract_indicators_by_position("工商银行", 2023, form="第一季度报告")
extract_indicators("贵州茅台", 2023, form="半年度报告")Concurrency
extract_indicators("工商银行", 2023, concurrency=4) # explicit cap
extract_indicators("工商银行", 2023, concurrency=1) # sequential
# Batch: multi-company concurrent extraction
extract_indicators_batch([("601398", 2023), ("600519", 2023)],
concurrency=2, extract_concurrency=4)
# → {"results": {"601398_2023": {...}, ...}, "failures": [...], "concurrency": 2}Section Cache
LLM responses are persisted to disk, keyed by (pdf_url, section_key, period, rules_hash).
Subsequent runs reuse cached records. Set LLM_SECTION_CACHE=off to disable.
Standalone CLIs
# Full engine extraction
python scripts/extract_indicators.py 601398 --year 2023 \
[--rules indicator_rules.json] [--extractor auto|llm|python] \
[--indicators 资本充足率,不良率] [--out-dir ./out]
# CSV-driven extraction
python scripts/extract_indicators_by_position.py 601398 --year 2023 \
[--csv docs/indicators_position.csv] [--extractor auto|llm|python] \
[--form 年度报告|半年度报告|第一季度报告|第三季度报告]
# Multi-year extraction
python scripts/extract_indicators_multiyear.py 601398 2023 2024
# Industry rules dashboard
python scripts/industry_rules_dashboard.py [port]
# or: fd-cn-report
# Generate industry rules from real reports
python scripts/generate_rules_from_real_reports.py
# Check industry coverage
python scripts/check_industry_coverage.py
# Seed industry rules
python scripts/seed_industry_rules.pyHK Stock Support
get_hk_company("00700") # → 腾讯控股
list_hk_filings("00700", year=2023) # → HKEX filings
get_hk_financials("00700") # → financial statements
get_hk_section("00700", year=2023, section="管理层讨论与分析")Official-website datasources (SSE / SZSE / BSE)
Direct, primary-source disclosure paths complementing CNINFO. Each exchange
client resolves a 6-digit ticker locally (no network) and lists disclosures
from the exchange's own site. BSE falls back to CNINFO when its own API is
thin, tagging each row with source: "bse" | "cninfo". Section extraction
reuses the same outline pipeline as CNINFO/HK.
# SSE (上交所) - 600/601/603/605/688/900 codes
get_sse_company("600519")
list_sse_filings("600519", year=2023) # -> sse.com.cn disclosures
get_sse_section("600519", year=2023, section="管理层讨论与分析")
get_sse_interaction("600519") # -> 上证e互动 Q&A
# SZSE (深交所) - 000/001/002/003/300/301 codes
get_szse_company("000001")
list_szse_filings("000001", year=2023) # -> szse.cn disclosures
get_szse_section("000001", year=2023, section="管理层讨论与分析")
get_szse_interaction("000001") # -> 互动易 Q&A (irm.cninfo.com.cn)
# BSE (北交所) - 430xxx / 83xxxx / 87xxxx / 88xxxx / 920xxx codes
get_bse_company("835185")
list_bse_filings("835185", year=2023) # -> BSE-native, else CNINFO fallback
get_bse_section("835185", year=2023, section="管理层讨论与分析") # result carries `source`Endpoints are undocumented. SSE (
query.sse.com.cn), SZSE (www.szse.cn/api), BSE (www.bse.cn), and the Q&A hosts (sns.sseinfo.com,irm.cninfo.com.cn) expose no official API contract and shift over time. Clients are thin, retry 429/5xx, and bypass proxy env (trust_env=False). RunCNREPORT_SELFCHECK_LIVE=1 uv run python selfcheck.pyto ping each endpoint - 4xx/5xx are flagged per source without failing the suite. Name-fragment resolution is not supported by these clients; resolve the name viaget_company(CNINFO) first, then pass the 6-digit code.
CSRC regulatory data (证监会)
CSRC regulatory data CNINFO/exchanges do not cover: announcements, IPO / 并购重组
review status, and enforcement. HTML-parsed with lxml; endpoints undocumented.
list_csrc_filings(begin_date="2024-01-01") # -> regulatory announcements
get_csrc_ipo_review("贵州茅台") # -> IPO review status row
get_csrc_merger_review("某科技股份公司") # -> M&A review status row
list_csrc_enforcement() # -> administrative-penalty actionsMinistry statistics (部级部门)
Structured economic/financial statistics from ministry-level departments,
complementing fd-cn-gov's catalog-archive scraping with data queries. NBS
has a JSON API; the others publish HTML tables. Results are TTL-cached under
.cache/stats/. Base URLs reuse fd-cn-gov's registry when importable.
list_ministries() # -> [{id, label, en, transport, base}, ...]
get_nbs_stat("A0201") # -> NBS GDP series {period: value} (dbcode="hgnd" annual)
get_ministry_stat("gacc") # -> GACC trade page parsed into HTML tables
get_ministry_stat("pboc", limit=20) # -> PBoC monetary tables
# Supported ids: nbs mof pboc safe gacc nfraMinistry stat-page paths and CSRC URLs are best-guess and undocumented - verify live via
CNREPORT_SELFCHECK_LIVE=1 uv run python selfcheck.py. Corrections are one-line edits inministry_stats_client._MINISTRIES/csrc_client._URLS.
Setup
uv sync # installs akshare, pypdf, fastmcp, ...
uv run python server.py # FastMCP over stdioSelf-check (no network):
uv run python selfcheck.py # DB + outline + company API + special reports
uv run python selfcheck_cache.py # report cache + three-statements extractionTests (offline):
uv run --with pytest python -m pytest test_cnreport.py -v -p no:logfireConfiguration
CNINFO and akshare are keyless. Other tools need env vars in .env:
Var | Used by | Required? |
|
| Yes for AI |
|
| Yes for ES |
| provenance writes, rules storage | Defaults to |
| report cache | Defaults to |
| user-visible PDF save directory | Optional |
| PDF upload to MinIO object store | Optional |
Architecture
fd-cn-report/
├── server.py # FastMCP server (@app.tool registrations)
├── cninfo_client.py # CNINFO API (keyless lookup + query)
├── hk_stock_client.py # HK stock API (akshare + HKEX)
├── financials_client.py # akshare financial statements (lazy)
├── cnreport_tools.py # Pure helpers (outline, extract, LLM, ES)
├── report_cache.py # On-disk PDF + text + outline cache
├── llm_section_cache.py # LLM response section cache
│
├── indicators_client.py # Rules engine (load, profile, route, extract)
├── indicators_extractors.py # Pluggable Python extractors
├── indicators_models.py # Pydantic extraction models
├── indicators_csv_migration.py # CSV → JSON rule migration
├── report_section_map.py # Section alias expansion + matching
│
├── rules_db.py # Rules database (SQLite via SQLAlchemy)
├── rules_models.py # Pydantic rule models
├── rules_skills.py # LLM rule generation + validation
├── cnreport_models.py # ORM models (LlmRule, ScriptRule)
├── cnreport_database.py # Database connection management
│
├── industry_taxonomy.py # 申万 L1 industry taxonomy
├── industry_coverage.py # Coverage checks for industry rules
│
├── docs/
│ ├── industry_taxonomy.json # 31 industry taxonomy
│ ├── industry_indicator_baseline.json # Baseline indicators per industry
│ ├── indicators_position.csv # Indicator catalog (CSV source)
│ └── indicators-methodology.md # Rendered methodology
│
├── scripts/
│ ├── industry_rules_dashboard.py # Web dashboard (CLI + MCP)
│ ├── generate_rules_from_real_reports.py # Per-industry rule generation
│ ├── generate_all_industry_rules.py # LLM rule generation
│ ├── seed_industry_rules.py # Universal rule seeding
│ ├── seed_missing_industry_rules.py # Industry-specific seeding
│ ├── extract_indicators.py # Standalone extraction CLI
│ ├── extract_indicators_by_position.py # CSV-driven extraction CLI
│ ├── extract_indicators_multiyear.py # Multi-year batch extraction
│ ├── check_industry_coverage.py # Coverage validation
│ ├── migrate_indicators_csv.py # CSV → DB migration
│ └── rules_dashboard.py # indicator_rules.json editor
│
└── .cache/reports/ # Downloaded PDFs + extracted text + outlinesAdding a New Industry
Add to
docs/industry_taxonomy.json:{"industry": "801xxx", "label": "行业名称", "report_kinds": ["annual-report", "interim-report", "quarterly-report"]}Add baseline indicators to
docs/industry_indicator_baseline.json:{"cn/801xxx/listed/annual-report": ["资产总计", "营业收入", "净利润", ...]}Add representative company to
scripts/generate_rules_from_real_reports.py:"801xxx": ("600xxx", "代表公司", "行业名称"),Define industry-specific sections:
"801xxx": [("section_name", "keyword", "keyword2", "guidance"), ...],Generate rules:
python scripts/generate_rules_from_real_reports.py --industry 801xxx
License
MIT
Available Tools
44 toolsai_extractC
Run LLM structured extraction over report section text.
Inputs: text (required), schema (required), prompt, max_chars. Returns: object
audit_rule_gapsB
Audit rule gaps from existing out/ bundles and write a stable JSON report.
Inputs: out_dir, max_files, output_path. Returns: object
clear_cacheC
Evict cached annual reports.
Inputs: year, stock_code. Returns: object
delete_indexA
Drop the cnreport-{year} Elasticsearch index and its metadata row.
Inputs: year (required), confirm. Returns: object
extract_indicatorsA
Extract many indicators for one company/year in a single pass.
Fetches the annual-report PDF once, groups report-rules by section (one LLM call per section), dispatches Python extractors individually, and computes derived ratios locally. Caches the bundle to disk so repeat calls are free.
Inputs: year (required), ticker_or_name (required), form, indicators, +1 more. Returns: object
extract_indicators_by_positionA
Extract the indicators named in a position CSV for one company/year.
Default CSV is docs/indicators_position.csv (override with csv_path).
Reads the CSV's indicator column, routes report/akshare/computed indicators
through the batch engine (one PDF fetch, batched LLM, python extractors,
computed ratios, bundle cache), and lists realtime/external indicators
(PE-TTM, PB, 市值, …) in skipped — they are not in the report PDF.
Inputs: year (required), ticker_or_name (required), csv_path, extractor, +1 more. Returns: object
extract_sectionC
Extract one section's body text by selector.
Inputs: source (required), selector (required), form, year, +3 more. Returns: object
get_bse_companyB
Resolve a BSE-listed company by 6-digit ticker (bse.cn).
Inputs: ticker_or_name (required). Returns: object
get_bse_sectionA
Extract a named section from a BSE company's annual-report PDF.
The filing may be served from BSE or CNINFO (fallback); the result carries
source. Returns: {stock_code, year, section, pdf_url, source,
outline_entry, text, char_count} or {error}.
Inputs: year (required), section (required), ticker_or_name (required). Returns: object
get_companyB
Resolve a CN-A-share company by 6-digit ticker or Chinese/English name fragment.
Inputs: ticker_or_name (required). Returns: object
get_csrc_ipo_reviewB
Query CSRC IPO (首发) application review status for a company.
Returns: {company, source, fields} (matching application row) or {company, source, error}.
Inputs: company_or_code (required). Returns: object
get_csrc_merger_reviewB
Query CSRC 并购重组 (M&A) review status for a company.
Returns: {company, source, fields} or {company, source, error}.
Inputs: company_or_code (required). Returns: object
get_filingB
Fetch one CNINFO announcement's metadata + PDF URL by id.
Inputs: announcement_id (required), ticker_or_name. Returns: object
get_financialsC
Return structured income/balance/cashflow statements for a CN-A-share company.
Inputs: ticker_or_name (required), period, statement. Returns: object
get_financial_statementsA
Extract the three major financial statements (三大报表) as text.
Resolves the company's filing PDF for (ticker, year, form) via the
report cache (no re-download on repeat), parses the table of contents,
and returns each statement's body text:
statements.income_statement(利润表) — prefers 合并利润表statements.balance_sheet(资产负债表) — prefers 合并资产负债表statements.cashflow(现金流量表) — prefers 合并现金流量表
Returns section text only — never PDF bytes. Statements not located
in the TOC are listed in missing (with the full available title
list so the caller can fall back to get_section with a custom selector).
Inputs: year (required), ticker_or_name (required), form. Returns: object
get_hk_companyB
Resolve a HK stock company by 5-digit ticker or name fragment.
Inputs: ticker_or_name (required). Returns: object
get_hk_financialsC
Return structured financial statements for a HK stock company.
Inputs: ticker_or_name (required). Returns: object
get_hk_sectionB
Extract a named section from a HK stock annual report PDF.
Inputs: year (required), section (required), ticker_or_name (required). Returns: object
get_indicatorA
Resolve one named financial indicator for a company + period.
Routes per the rule's source_type: akshare line items, annual-report PDF
section (LLM or registered Python extractor), or locally-computed ratio.
Inputs: year (required), indicator (required), ticker_or_name (required), period. Returns: object
get_ministry_statC
Fetch a ministry's statistics page and parse its HTML tables.
Inputs: ministry_id (required), url, limit. Returns: object
get_nbs_statB
Query an NBS (国家统计局) macro statistic by indicator code.
Inputs: indicator_code (required), dbcode. Returns: object
get_sectionA
Resolve a company's filing PDF and extract one named section.
Convenience wrapper: (ticker, year, section, form) → CNINFO lookup → PDF URL → existing outline-extraction pipeline.
Inputs: year (required), section (required), ticker_or_name (required), form. Returns: object
get_special_reportB
Retrieve a special-type report for a CN-A-share company by CNINFO category.
Inputs: category (required), ticker_or_name (required), year, limit, +1 more. Returns: object
get_sse_companyB
Resolve a SSE-listed company by 6-digit ticker (sse.com.cn).
Inputs: ticker_or_name (required). Returns: object
get_sse_interactionB
Fetch 上证e互动 investor Q&A for a SSE-listed company (sns.sseinfo.com).
Inputs: ticker_or_name (required), limit. Returns: object
get_sse_sectionC
Extract a named section from a SSE company's annual-report PDF.
Inputs: year (required), section (required), ticker_or_name (required). Returns: object
get_szse_companyA
Resolve a SZSE-listed company by 6-digit ticker (szse.cn).
Inputs: ticker_or_name (required). Returns: object
get_szse_interactionB
Fetch 互动易 investor Q&A for a SZSE-listed company (irm.cninfo.com.cn).
Returns: {stock_code, source, questions: [...]} or {error}.
Inputs: ticker_or_name (required), limit. Returns: object
get_szse_sectionC
Extract a named section from a SZSE company's annual-report PDF.
Returns: {stock_code, year, section, pdf_url, outline_entry, text, char_count} or {error}.
Inputs: year (required), section (required), ticker_or_name (required). Returns: object
index_recordsC
Bulk-index extracted records into cnreport-{year}.
Inputs: year (required), records (required), report_id (required), section_id (required), +3 more. Returns: object
list_bse_filingsB
List a BSE-listed company's disclosures (BSE-native, CNINFO fallback).
Each row carries source: "bse" | "cninfo". Returns: {filings, count} or {error}.
Inputs: ticker_or_name (required), year, limit, title. Returns: object
list_cacheA
List cached annual reports.
Each entry carries the parsed provenance (stock_code / year /
form / announcement_id for convenience-tool fetches, or
kind: "url" for raw extract_section/list_outline URL fetches),
cached_at (file mtime, ISO-8601 UTC), and size (sum of its
.pdf + .txt + .outline.json).
Inputs: no named parameters. Returns: object
list_csrc_enforcementC
List CSRC administrative-penalty / enforcement actions (csrc.gov.cn).
Returns: {filings, count} or {error}.
Inputs: limit, begin_date. Returns: object
list_csrc_filingsC
List CSRC regulatory announcements/notices (csrc.gov.cn).
Inputs: limit, category, end_date, begin_date. Returns: object
list_filingsC
List a CN-A-share company's CNINFO disclosures.
Inputs: ticker_or_name (required), form, year, limit, +1 more. Returns: object
list_hk_filingsC
List a HK stock company's filings/announcements.
Inputs: ticker_or_name (required), form, year, limit. Returns: object
list_indicatorsC
Browse the banking-indicator rule set (data-driven by indicator_rules.json).
Inputs: query, module, company. Returns: object
list_ministriesA
List the supported ministry statistics sources.
Returns: {ministries: [{id, label, en, transport, base}], count} or {error}.
Inputs: no named parameters. Returns: object
list_outlineC
Fetch a Chinese annual report and return its 目录 outline.
Inputs: source (required), fetcher. Returns: object
list_report_typesB
Browse the CNINFO disclosure category catalog.
Inputs: group. Returns: object
list_sse_filingsB
List a SSE-listed company's disclosures directly from sse.com.cn.
Inputs: ticker_or_name (required), year, limit, title. Returns: object
list_szse_filingsA
List a SZSE-listed company's disclosures directly from szse.cn.
Returns: {filings, count} or {error}.
Inputs: ticker_or_name (required), year, limit, title. Returns: object
open_industry_rules_dashboardA
Start the industry rules web dashboard with filtering and search.
Opens a web UI at http://localhost: showing all 21,698 LLM rules across 31 申万 L1 industries. Filter by industry, module, keyword; sort by any column; paginated view.
Inputs: port. Returns: object
search_reportsC
Full-text + filtered search over cnreport indices with highlights.
Inputs: query (required), size, year, from_, +3 more. Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 44 tools
Many tools are differentiated only by exchange or source (e.g., get_company vs get_szse_company, list_filings vs list_szse_filings), and multiple extraction/financial tools overlap in purpose (get_section, extract_section, get_financial_statements, get_financials). Descriptions are detailed and clarify the intended scope, but the sheer volume makes misselection likely for an unfamiliar agent.
Nearly all 44 tools follow a consistent snake_case verb_noun pattern (list_filings, get_section, extract_indicators). Minor deviations like ai_extract are rare and do not disrupt the overall predictability.
44 tools far exceeds the ideal range and feels excessive even for a broad server spanning multiple exchanges, data types, extraction, indexing, caching, and dashboards. The count risks overwhelming agents despite each tool having a distinct niche.
The surface covers discovery, retrieval, extraction, indexing, and caching for CN/HK financial disclosures across many sources. Minor gaps exist, such as no indicator listing for NBS or no specific filing-fetch for some exchanges, but these are workable with external knowledge.
Maintenance
Related MCP Connectors
Financial data and research MCP for US/CN/JP equities: filings, statements, ownership, signals.
China A-share & equity research MCP server (回声投研) for market quotes, filings, financials, fund flows, concepts & sentiment.
China A-share market data over MCP: 22 tools for quotes, K-line, financials, money flow, top-trader boards, sectors, macro, convertible bonds and factor screening. Five tools need no API key, so you can connect and try it immediately.
75 MCP tools: SEC financials, FRED economics, IRS 990, FDA, FX, UK Companies House.
Related MCP Servers
- AlicenseAqualityDmaintenanceMCP server for Chinese A-share (Shanghai/Shenzhen/Beijing) financial statements: pull annual reports, run 4 industry-aware accounting cross-checks, and peer-compare companies with derived avg-equity ROE.36MIT
- FlicenseNot gradedqualityDmaintenanceMCP server that wraps SFC financial data API into 32 tools for comprehensive A-share market data, including real-time quotes, rankings, limit-up statistics, news, themes, financials, charts, research reports, and watchlists.-
- FlicenseNot gradedqualityDmaintenanceAn MCP server that provides tools and resources to access company financial filings, industry classifications, and related data via the Financial Reports API.7-
- AlicenseNot gradedqualityDmaintenanceMCP server for analyzing SEC filings (10-K, 10-Q, 8-K) with industry-aware financial extraction and BERT-based NLP.1MIT