> ## Documentation Index > Fetch the complete documentation index at: https://onetest.ai/llms.txt > Use this file to discover all available pages before exploring further. # How OneTest Builds > The principles behind OneTest's Applied AI projects # Tools for shared work between people and agents OneTest builds Applied AI products around a simple idea: AI becomes useful when it works from durable context, produces inspectable artifacts, and participates in the same systems as the people around it. ## Our principles ### Keep the work visible Plans, test assets, presentations, decisions, and reports should remain inspectable after an agent session ends. Whenever practical, they belong in files and repositories that teams already understand. ### Give agents explicit contracts An instruction is not enough. Useful agent workflows need acceptance criteria, structured tools, observable outcomes, and completion gates. ### Fit existing workflows Our products integrate with Git, GitHub, VS Code, browsers, desktop applications, MCP, and standard command-line tools. Adoption should not require moving all work into a proprietary environment. ### Prefer focused tools OneTest is a collection of focused projects rather than a single monolithic platform. Each tool should solve a concrete problem well and compose with the others through open interfaces. ### Build in public We publish source code, design decisions, limitations, and unfinished experiments. Public development makes the work easier to inspect, challenge, reuse, and improve. ## Our roots OneTest started with AI-assisted testing. That experience produced a broader set of questions: * How should agents share plans with people? * How does work survive beyond a chat session? * How can outcomes be verified rather than merely claimed? * How should agent-created artifacts fit established engineering workflows? OneTest TMS continues the software-quality work. DeepSeek Harness Desktop, Octodeck, and Octobots explore the same questions across desktop AI, presentations, and project execution. See the current OneTest product family. Follow development, open an issue, or contribute code and ideas.