Add live weather, news, or stock data to an AI agent workflow without paying for commercial data APIs.
Build a daily digest pipeline that automatically pulls crypto news and currency rates from multiple free sources.
Let an AI agent discover relevant free public APIs by category and keyword instead of hard-coding every URL.
Free API Discovery is a Python tool designed to sit between a natural-language request and the data that request is asking for. The README is written in Chinese. The project is described as a "Codex Skill," meaning it is intended to be called by an AI agent or automation system rather than used directly by a human typing commands. The tool handles two kinds of requests. When someone asks for a live result, such as the current weather in Shanghai, the latest cryptocurrency news, a currency conversion, or a stock snapshot, the tool routes that request to a curated set of free public data sources and returns a structured JSON result. It currently supports ten categories of live data: general news, crypto news, financial news, weather, geographic lookups, country profiles, currency exchange rates, quick encyclopedia lookups, cryptocurrency market snapshots, and individual stock context. News results are returned with both the original title and a Chinese summary. When someone asks for sources rather than results, such as finding free GraphQL APIs for market data or listing all available API categories, the tool searches a local index built from two large public API directories. It filters candidates by category, authentication requirements, and keyword relevance, then returns a short list of matching APIs rather than dumping the entire directory into an AI's context. This keeps token usage low when the tool is called repeatedly inside an automated workflow. The design is intentionally conservative about reliability. Free public data sources go down or return empty results unpredictably, so the tool uses multiple sources in parallel for news, has fallback sources for market data, and reports how many sources actually responded in each result rather than presenting results as uniformly authoritative. The tool is not a hosted service and not a comprehensive data platform. It is a local-first routing layer suited to lightweight automation, daily digest pipelines, or agent workflows that need to query public data cheaply and repeatedly. No license file is included in the repository.
← wearescientist on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.