Sealshot: Free Open-Source macOS App Treats Screenshots as Searchable Documents

·1 views

A small development team has released Sealshot, a free and open-source screenshot application for macOS. The app was created to address the problem of screenshots becoming hard to find, reuse, or accidentally exposing sensitive data like API keys or customer information. Sealshot offers features including region and scrolling capture, OCR-powered search, editable annotations, and sensitive information detection before sharing. All processing is handled locally on the device, with encrypted local storage and metadata generation. The team is actively seeking feedback from developers, QA engineers, support staff, and security professionals.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Developer shares five hard-learned lessons for deploying static sites to Cloudflare Workers

A developer documented the setup and pitfalls encountered while migrating a static site from a hosted blogging platform to Cloudflare Workers, using GitHub Actions for CI/CD. The pipeline builds the site, checks for dead links and bad metadata, runs Lighthouse audits, and only deploys to production after all quality gates pass. A key confusion arose from Cloudflare offering two similar products — Pages and Workers — with the latter now being the actively developed option that supports redirects, custom headers, and custom domains. The author also warns that using a custom API token instead of Cloudflare's built-in 'Edit Cloudflare Workers' template leads to cryptic authentication errors, and recommends verifying token permissions with 'wrangler whoami' before connecting to CI. Additionally, once a wrangler.jsonc config file is added, any omitted settings are silently disabled, causing staging URLs and preview links to disappear without obvious error messages.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How Startups Should Choose an AI API Router for Property Moderation Tasks

For property-management startups using AI moderation, selecting the right single-key API router requires more than comparing raw token prices across providers like OpenAI, Claude, and Gemini. The true cost metric is cost per accepted classification, which accounts for inference fees, retries, invalid outputs, and operational labor. Developers are advised to build a frozen replay set of around 500 diverse moderation reports to benchmark each router candidate under identical conditions. Tools like OpenRouter, Portkey, and LiteLLM each serve different operational needs, ranging from managed routing to self-hosted gateways and direct adapters. The core decision revolves around portability and policy control, not simply finding the lowest per-token price.

0 comments Read more at DEV Community

ProgrammingHacker News ·

How Fluorescent Lamps Interact With Sound: A Technical Exploration

A technical blog post published on coredump.cx examines the relationship between fluorescent lamps and sound or audio signals. The article, shared on Hacker News, explores whether fluorescent lighting systems can pick up or respond to acoustic input. The post received 3 points on Hacker News at the time of submission. No community comments had been posted on the thread. The full technical details are available via the original blog post link.

0 comments Read more at Hacker News

ProgrammingDEV Community ·

OpenAI Integrates ChatGPT with Electronic Health Records for Clinical Use

OpenAI is enabling ChatGPT to connect directly with electronic health records and other health-system data sources in supported deployments. The move is aimed at embedding AI assistance within existing clinical workflows, reducing the need for clinicians to switch between separate tools and patient charts. The integration is part of OpenAI's broader ChatGPT for Healthcare initiative, which includes HIPAA-compliant workspaces and responses drawn from trusted medical sources. OpenAI states that data connected through Health features can be configured to avoid model training and will be deleted from its systems within 30 days. Exact availability, supported EHR vendors, and pricing remain dependent on individual enterprise partnerships and deployment arrangements.

0 comments Read more at DEV Community