面向開發者與 AI Agent 的 API 實戰手冊
教學、橫向對比與用例，協助你為 LLM 回答找出處、搭 Agent 流程、把 AI 功能推到正式環境。
Google Patents API Is Dead — 6 Alternatives for Prior-Art and Freedom-to-Operate Search
Google Patents Public Datasets retired without a documented replacement, and the USPTO migrated PEDS into a new Open Data Portal mid-2026, breaking pipelines on the way. Here's what actually works in 2026 for prior-art, FTO, and IP-research workloads — six alternatives, side by side.
How to Build a Scientific Literature-Review Agent Without Getting Rate-Limited
Build a literature-review agent on raw arXiv + PubMed + Semantic Scholar today and you'll hit 429s before you finish ten papers. Here's why the rate limits got worse, what PaperQA / Undermind actually do under the hood, and a working pattern that survives a real review session.
Polymarket vs Kalshi API: A Developer's Side-by-Side Guide (Auth, CLOB, WebSocket, Historical Data)
Polymarket and Kalshi run the same primitive — yes/no contracts on a CLOB — through completely different APIs. One demands EIP-712 signatures and a Polygon wallet; the other is a REST endpoint with optional FIX. If you're building a forecasting agent, an arbitrage bot, or a smart-money monitor, here's the side-by-side guide that should already exist.
Citation-Grounded UK Case Law Retrieval After Ayinde v Haringey
Ayinde v Haringey changed the calculus for UK legal AI overnight. A barrister cited five fake cases generated by an LLM and got referred to the Bar Standards Board. Hallucinated citations now have professional-conduct consequences. Here's the developer's guide to building citation-grounded UK case law retrieval — TNA Find Case Law, legislation.gov.uk, neutral-citation parsing, and how to wire it into a working agent.
Wiring ClinicalTrials.gov v2 + openFDA + ChEMBL into a License-Clean Drug Intelligence Endpoint
Pharma R&D, medical-AI startups, and pharmacovigilance teams all want the same thing: one endpoint that pulls trials, labels, adverse events, and bioactivity in a license-clean way. Here's the working architecture, with the trap doors that have surprised teams in production.
Build a Research Agent with Search + URL Extract (Claude tool use, end-to-end)
Most 'AI research agent' tutorials stop at 'here's a tool definition.' This one ships a working agent: question in, cited answer out. Search, extract, reason, cite — all under 120 lines of Python.
Building a Due-Diligence Agent on SEC Filings (10-K, 10-Q, 8-K, Earnings)
Reading a 10-K is mostly Ctrl+F. Doing it for fifty companies is a job. Replace the boring 80% with a search-and-extract agent against SEC EDGAR — keep the 20% that matters for the human analyst.
Why APIs Should Charge Only on Success — The Case for HTTP-200 Billing
Most APIs charge on every billable call. AI agents retry on flaky upstream services constantly — which means the legacy 'always charge' model effectively taxes resilience. Here's the case for billing only on HTTP 200, and what it changes about how you design agents.
Tavily、Exa、Serper、API Pick 對比：給 LLM 用的網頁搜尋 API 怎麼選？
Tavily、Exa、Serper、API Pick 都自稱是「給 LLM 用的搜尋層」。它們在回傳結構、過濾維度與計費方式上差別其實不小。下面是來自第一線工程師的橫向對比。
Firecrawl vs Jina Reader vs API Pick：URL 內容擷取 API 怎麼選
凡是寫過「讓 Agent 讀一組 URL 並摘要」的人，都被 URL 清洗那一步坑過。Firecrawl、Jina Reader、API Pick Extract 各有解法，下面是實作對比。
4 步給 OpenAI / Claude Agent 接入即時網頁搜尋
想讓 Agent 的回答基於最新資訊？只要兩樣東西：一個回傳 LLM 友善 snippet 的搜尋 API，加一個模型能呼叫的 tool 定義。本文把這兩件事拆成 4 步。
用 News Search API 做一個早間簡報 Agent
創業者、PM、分析師早上 8 點想要的都是同一樣東西：一份精煉的「昨夜發生了什麼」摘要。一個 News Search API、一次 LLM 呼叫、一份 cron，下午就能交付。