TOOL API

tiktok_ads_creative_testing_agent

Any agent can call this tool when the user expresses creative testing intent, such as wanting to test creatives, understand creative performance, ensure specific creatives receive a specified amount of spend or impressions, or compare different creatives. It can also be called when the user wants to manage or check existing creative test plans. This tool supports creative-level testing within existing TikTok Ads ad groups. It identifies eligible ad groups and creatives, collects missing inputs, creates or updates test plans to give each test creative sufficient exploration with specified spend or impression targets, and returns interim and final test reports. Notice: Display agent's full response, including any web page and video links attached to the text, without shortening the content. Video links in particular matter, because advertisers rely on them to judge which video it is. Expected inputs are the TikTok Ads advertiser account ID and the user's original request in natural language. No execution before explicit user confirmation. After the user confirms execution, the tool will return a plan_name, which the user can use to manage the progress of a specific test (e.g., "Stop the dry_run creative test plan."). Session and turn handling (read carefully — this tool does not keep state for you): This tool is stateless between calls. YOU (the calling agent) own the conversation's session_id. The first time a user starts a creative testing conversation, generate a new random 6-digit session_id. For every follow-up turn that belongs to the SAME creative testing conversation (more suggestions, further explanation, a different time period, a new benchmark, or a correction to the creative testing request), reuse the same session_id as context so continuity is preserved. Start a new session_id only when the user begins an unrelated, new conversation.