wtf is twelvelabs-playground?

twelvelabs-playground is a small command line tool for trying out TwelveLabs, a video understanding service, without writing any code. TwelveLabs can index a video and then let you search within it using plain language, or ask it direct questions about what happens in the footage. Setting it up means creating a Python virtual environment, installing the listed dependencies, and pasting an API key from the TwelveLabs playground website into a local .env file. The README notes that the free tier includes 10 hours of video indexing with no credit card required, though a separate notes section says the free tier actually allows 600 minutes, so the exact limit may vary. Once set up, you create an index one time and save its ID, then upload videos either from a local file or a public URL, which takes a few minutes to process. After a video is indexed, you can list what has been indexed, run a semantic search across every indexed video using a natural language description of a moment, or generate a summary, a chapter breakdown, or a set of highlights for a specific video. You can also ask an indexed video open ended questions, such as listing the key claims made along with their timestamps. There is a separate quick mode powered by a model called Pegasus 1.5 that skips indexing entirely: you can point it at a video URL up to two hours long, or a local file under 36 megabytes, and ask a question directly, useful for one-off analysis without committing to a full index. The README suggests uses like turning a long conference talk into a chaptered table of contents, finding a visual moment you remember but cannot locate, or reviewing exercise form from a workout video. Videos and the .env file are excluded from the repository by design, and the notes point out that indexing cost is the main expense to watch as usage grows. No software license is mentioned in the README.