Urgent.News

What's breaking now, across thousands of outlets.

Tech

Monitor UK Companies House changes as delta events — without building your own polling pipeline

If you build anything that touches UK companies — credit checks, KYB onboarding, supplier risk, B2B lead scoring — you eventually need to answer the same question continuously: what changed at the companies I care about? The official Companies House public data API is free and well-documented. But it's a request/response register, not a feed. To watch a portfolio of companies you have to build…

Companies House, the public register of UK companies, provides a free API for accessing company data. However, the API is a request/response system, not a real-time feed, requiring developers to build their own polling mechanism to track changes in company information. This involves storing snapshots of each company, comparing fields, and determining if changes warrant an alert. This process is time-consuming and requires ongoing maintenance.

To simplify this process, Steadfetch offers a hosted delta feed called companies-house-monitor. This service allows users to maintain a private watchlist and receive only the changes made to the companies on that list, in the form of validated JSON events. This approach eliminates the need for developers to build and maintain their own polling pipeline.

Each delta event contains a stable schema with information such as the company number, name, status, risk indicators, and a human-readable description of the risk. The events are sourced from public register data, making them compliant with the Open Government Licence v3.0. The service supports two types of watches: company watches, which track a single register record, and query watches, which are saved searches that can return multiple new companies.

To use the service, developers can install the Apify CLI, add their first watch, and schedule daily runs of the monitor. The first run delivers a baseline, and subsequent runs only deliver changes. These changes can be easily consumed by JavaScript or Python scripts. The Actor is priced based on the number of delta events, monitor checks, and run starts, with a free monthly credit for the Apify free plan.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

How to Learn React Hooks (and Stop Fighting useEffect)

The first React app I wrote with hooks had a useEffect that fetched a user profile. It worked on the first load. Then I added the user ID to the dependency array because the linter told me to, and the…

  • Start by understanding what a render is in React hooks.
  • Use the updater form of useState for dependent state values.
  • Properly use the dependency array in useEffect to avoid stale data.

Ofrenda: A Remembrance Builder That Never Leaves Your Laptop (clear, and states the open-source angle right away)

What I Built Every family has a few people whose stories only exist in someone’s head. A grandmother who hummed while she cooked. An uncle with one phrase he said at every dinner.

  • Ofrenda is an open-source tool for creating personalized remembrances.
  • Users input details about loved ones, which a local Gemma model uses to generate drafts.
  • Completed remembrances are exported as standalone HTML files with photos and animated candles.

Fintech Ticket Triage with Node.js RAG: PDF Upload, Semantic Search, and Citations

Short answer: use Node.js for the upload workflow, pgvector for metadata-aware semantic search, and embeddings plus citations as a validated contract; choose a shared OpenAI-compatible runtime only…

  • PDF uploads divided into overlapping chunks for semantic search
  • Embedded chunks with filename, page, and section metadata
  • Generated answers include citations and provenance for verification

Day 15 of 30: the halfway map, fourteen ideas worth keeping

Halfway through the series. Instead of a self-congratulating check-in, here's something more useful: a map of the fourteen ideas so far, each in a paragraph, so you can skim the series in five minutes…

  • Day 15 marks halfway point in 30-day series
  • Focus on fourteen ideas for safe defaults and systems
  • Next topics: Three.js, Row Level Security, test plan

Tool Selection at Scale: What Top-K RAG Leaves Out

Embed, retrieve, let the LLM decide works for about 100 tools. At 1,000 it needs a control plane: hybrid retrieval, hard filters, a learned reranker, progressive disclosure and per-stage evals.

  • Tools have side effects beyond simple category classification
  • Retrieval controls context and improves accuracy
  • Recall chain is product of conditional recalls at each stage

Why I Separated MIB Compilation from SNMP Execution in Python

Why I Separated MIB Compilation from SNMP Execution in Python When I began building network-management tools, I wasn't planning to write a MIB compiler or implement an SNMP runtime.

  • Author separated MIB compilation and SNMP execution into distinct components
  • Chose JSON as primary MIB artifact for stable, portable data contract
  • Created Trishul SNMP, asynchronous manager independent of Trishul SMI

More from Sunday 1 November →