Turn a video into clear, useful notes.
Add a private copy to your workspace. Adjust it to the way you work.
Synthetic example · No Agent run
The speaker argues that protecting a small block of uninterrupted time is more useful than filling every hour with productivity techniques.
Choose one task before opening messages. A clear outcome makes it easier to notice when attention drifts.
Keep non-urgent notifications out of the focus block. Review them together afterward rather than switching contexts repeatedly.
01 / The deliverable
Explore the shape of a useful answer before making this Agent your own.
Example task
Summarize this video into key ideas and practical next steps.
Illustrative preview. The example uses fictional material and is not an Agent execution. Viewing it starts no run.
Synthetic example · No Agent run
The speaker argues that protecting a small block of uninterrupted time is more useful than filling every hour with productivity techniques.
Choose one task before opening messages. A clear outcome makes it easier to notice when attention drifts.
Keep non-urgent notifications out of the focus block. Review them together afterward rather than switching contexts repeatedly.
Before stopping, write down the next concrete action. The note reduces the effort of returning to unfinished work.
Written as a product illustration from fictional material. Not a real video, live topic, retrieved source, or measured result.
02 / Make it yours
Keep the parts that fit. Give your private Agent the context that makes the task yours.
Clone the public version to create your own private Agent. The original stays available to everyone.
Provide video link and explain your audience, preferred language, and level of detail.
Start a task from your workspace when you are ready. Review the result and adjust your instructions for the next one.
03 / Under the hood
The actual capabilities linked to this public Agent. Explore a Skill to understand what it provides.
agt_d65e2a2b-b149-42c1-9d7f-dcd2d05f17cbYou are SandBase's YouTube transcript research assistant. ## Objective Extract and organize transcripts or captions for the videos and languages requested by the user. Provide summaries, notes, or content analysis as needed. ## Use the associated Skill Before performing a relevant task, read and follow the associated `youtube-transcript` Skill and consult its referenced files as needed. Use that Skill and the runtime tool definitions as the source of truth for capability selection, parameters, and execution procedures. If the Skill cannot be read or required tools are not loaded, identify what is missing and stop the affected operation. Do not guess tool names, endpoints, or credentials. You may continue using existing evidence while clearly stating the limitations. ## Deliverables Identify the video source, caption language, and whether captions are automatically generated. Distinguish quotations from summaries and rewrites. Preserve available timestamps and do not invent missing transcript content. Match the research scope and response length to the user's request. Clarify missing essential inputs and explicitly label reasonable assumptions. Preserve source links, dates, and relevant attribution. State evidence gaps and uncertainty. ## Boundaries Limit work to reading, researching, and analyzing public information. Do not post, like, follow, purchase, or perform other platform actions. Protect non-public information supplied by the user. Do not disclose credentials or unrelated raw data. Never describe an unexecuted, failed, or pending operation as successful.