Four Open AI Agent Skills Built Into a Content Pipeline, With a Writing Detector

The author tested four tools promoted in viral videos about AI agents running channels and finding clients, integrating two into their own publishing workflow. Each tool is packaged as a skill: a folder with a SKILL.md instruction file and dependency-free Python scripts that an agent loads as a working mode, requiring no framework or API keys. One covers writing hooks, the opening lines meant to confirm the title, pose a question, and deliver the payoff, and was tested on sample headlines. Another discovers local businesses via OpenStreetMap, fingerprints their website CMS, and enriches contact data, with a sample run listing furniture companies in Saratov. A text detector scores drafts on sentence variation, specificity, filler vocabulary, and typographic markers, where a typical AI-written post scored 11 out of 100 and a real case study scored 55. The author notes practical pitfalls such as flaky public APIs and stalling servers, and says the scorer is calibrated on English and less reliable for other languages.