bigdata-skill

Pull structured financials, prices, and sentiment data from Bigdata.com via SDK and REST endpoints.

1.4k|216|Updated Oct 22, 2025
One-click install
npx skills add https://github.com/daymade/claude-code-skills --skill bigdata-skill
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: bigdata-skill
Source: https://github.com/daymade/claude-code-skills/tree/main/daymade-financial/bigdata-skill
Command: npx skills add https://github.com/daymade/claude-code-skills --skill bigdata-skill

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires bigdata-client, requests, and includes scripts (resource) and references (resource) components.

What problem does it solve?

The Bigdata.com MCP server returns only pre-synthesized prose and tearsheets, hiding the machine-readable substrate — per-chunk sentiment, entity spans, fiscal-period time series, and universe screeners — needed to build investment-research data pipelines.

Core Features & Use Cases

  • Structured Financial Data via REST: Access analyst estimates, earnings surprises, event calendars, ratings, price targets, full financial statements, TTM ratios, daily prices, dividends, and revenue segments through official /v1/* endpoints the SDK never wrapped.
  • Annotated News Search: Retrieve news/filing/transcript chunks with numeric sentiment scores and entity character spans that the MCP strips out.
  • Cost-Guarded Operations: Built-in CostTracker and CostModel measure chunk-level billing, veto over-budget backfills before running, and enforce ChunkLimit to avoid the doc-limit billing trap.
  • Use Case: Resolve NVIDIA to its rp_entity_id, pull forward analyst consensus and the earnings calendar for free, then run a cost-capped annotated chunk search on "data center" keyword — all with measured quota spend.

Quick Start

Set the BIGDATA_API_KEY environment variable, install bigdata-client in a Python 3.12 venv, then ask the assistant to pull analyst estimates and sentiment data for a ticker using the bigdata toolkit.

Frequently Asked Questions about bigdata-skill

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I get structured financial data from Bigdata.com API?

Use the bigdata-client SDK's internal bd._api.http transport to POST to official /v1/* endpoints like analyst-estimates/query, income-statement/query, and price/daily/query. These endpoints reuse the same JWT authentication and are free of chunk billing.

Why does the Bigdata MCP not return sentiment scores per news chunk?

The Bigdata MCP server is chat-optimized and returns only synthesized prose and tearsheets, stripping per-chunk sentiment numbers and entity character spans. Drop to the SDK's AnnotatedSearcher or REST layer to access the underlying structured fields.

Does Bigdata.com work for Chinese A-share stocks?

Structured financial data works for A-shares when you resolve entities via English names or ISIN codes, since Chinese name lookups return zero hits. However, Chinese-language NLP (entity detection and chunk sentiment) is a data-source-level dead end.

How does Bigdata.com chunk billing work?

One query_unit equals 10 chunks, and only chunk-search operations are billed — structured endpoints like financials, prices, and estimates are free. Always pass ChunkLimit(n) instead of a bare integer to Search.run, since document limits bill by the full chunk page.

Why do I get SSL UNEXPECTED_EOF errors on the first Bigdata API call?

The SDK's HTTP layer does not retry SSL handshake EOF errors, so a single network blip becomes a hard exception, especially through outbound proxies. Wrap every network call in a retry helper that catches transient SSL, EOF, and connection markers.