SyncGTM now works inside Claude and ChatGPT. Learn more
All GTM agents

TikTok Search Query

/tiktok-search-query

Searches TikTok by topic, hashtag or keyword and returns the ranking videos with caption, author handle and video URL, for a flat 0.4 credits per call.

Download Skilltiktok-search-query.zip · free

Overview

`tiktok_search_query` is a SyncGTM MCP tool that takes a topic, a hashtag or a plain keyword and returns the TikTok videos ranking for it, each with its caption, its author handle and its video URL. Cost is a flat 0.4 credits per call, so 100 results cost exactly what 10 cost. Use it to find the accounts and formats winning a topic before you commit to producing for it, and to collect the video URLs that feed `tiktok_comments_from_post`.

What it does

  1. 1

    Takes one required input: `query` — a topic like "cold email tips", a hashtag like #b2bsales, or a plain keyword. One topic per call; a mixed string returns a worse set for both.

  2. 2

    `max_results` controls how many videos come back — 1 to 100, default 10. It does not change the price, so leave it at 50 for research and drop to 10 only for a spot check. `query` and `max_results` are the only two parameters: there is no date, region, language or sort filter.

  3. 3

    Returns a `results` array. Each item carries exactly three fields — `description` (the caption), `author` (the handle) and `url` (the video link you can hand to another tool).

  4. 4

    Searches across accounts you have not identified yet. When you already know the handle, `tiktok_profile_video` (0.5 credits) is the direct route to that account's feed — it takes `username` or `profile_url` plus `max_posts`, not `max_results`.

  5. 5

    Returns no view counts, like counts, follower counts, emails, phone numbers or company names. Rank accounts by how often they repeat in the results, not by engagement you do not have.

  6. 6

    Costs 0.4 credits per call, flat. The multiplier is call count — hashtag plus keyword across 12 topics is 24 calls and 9.6 credits.

How to use it

  1. 1

    Download the skill file — it saves as tiktok-search-query.zip.

  2. 2

    In Claude, open Settings → Capabilities → Skills and upload the .zip (or unzip the tiktok-search-query/ folder into .claude/skills/ for Claude Code).

  3. 3

    Connect the SyncGTM MCP server using the setup guide for your client. It is browser sign-in, no API key, and no TikTok account is needed.

  4. 4

    Type /tiktok-search-query and give it one topic, hashtag or keyword plus the number of results you want, for example: search TikTok for "cold email tips", 50 results.

  5. 5

    The one setting worth tuning is `max_results`. Cost is flat, so raise it to 50 or 100 for any real research pass and run the hashtag form and the keyword form as two separate calls — they surface different videos.

Use cases

Topic saturation check

Search a topic at 50 results before you brief a video series. Read the `description` captions to see which angles already repeat and which one nobody has taken.

Creator and partner shortlist

Search your category hashtag and count which `author` handles appear more than once. Those repeat accounts are your affiliate, partnership and paid-collab shortlist.

Competitor content monitor

Search a competitor's brand name weekly, then your own, and compare video counts and unique accounts. It is a share-of-voice read that costs 0.8 credits a week.

Sourcing videos for comment mining

Use search purely to collect `url` values on a buyer pain point, then pass each one as `post_url` to `tiktok_comments_from_post` to pull the objection language underneath.

GTM workflow examples

Build a creator partnership shortlist for a category

  1. 1Call `tiktok_search_query` with `query` "#b2bsales" and `max_results` 100 to collect 100 videos, each with `author`, `description` and `url`.
  2. 2Call `tiktok_search_query` again with the plain keyword "b2b sales" and `max_results` 100, then dedupe both sets by `url` — hashtag and keyword return different accounts. Two calls, 0.8 credits.
  3. 3Count results per `author`, take the 10 handles that appear most, and call `tiktok_profile_video` on each with `username` and `max_posts` 50 to confirm they post weekly rather than once (0.5 credits per handle, 5 credits).
  4. 4For the handles whose captions or bio name an employer, call `find_people` with `current_company_names` set to that company, `current_titles` such as ["Founder", "Head of Marketing"] and `limit` 5 to find the LinkedIn profile — it bills 0.3 credits per profile returned, so always pass `limit`.
  5. 5Call `find_work_email` on the matched LinkedIn URLs only (1 credit each) to make the shortlist contactable.

Outcome: A ranked 10-creator shortlist with post cadence, dominant format, and a LinkedIn profile plus work email for every creator whose employer is named publicly.

Mine buyer objection language before writing the campaign

  1. 1Call `tiktok_search_query` with `query` "cold email tools" and `max_results` 50 to get the 50 videos ranking for the category.
  2. 2Call `tiktok_search_query` with `query` "#coldemail" and `max_results` 50, then merge and dedupe by `url` to widen the pool to roughly 80 unique videos. Two calls, 0.8 credits.
  3. 3Pick the 5 videos whose `description` names a tool or a specific pain, then call `tiktok_comments_from_post` on each with `post_url` and `max_results` 100 (0.4 credits per video, 2 credits) to pull up to 500 comments.
  4. 4Call `tiktok_profile_video` with `username` and `max_posts` 50 on the two handles those videos belong to (0.5 credits each, 1 credit) to see which hooks they repeat across their feed.

Outcome: An objection swipe file in verbatim buyer wording, ranked by how often each objection repeats and by comment `likes`, plus the hooks those two creators keep reusing — 3.8 credits for the whole pass.

Monday competitor share-of-voice snapshot

  1. 1Call `tiktok_search_query` with the competitor's brand name as `query` and `max_results` 50 to list every video mentioning them and who posted it.
  2. 2Call `tiktok_search_query` with your own brand name and `max_results` 50 for the same read — the difference in video count and unique `author` values is the share-of-voice gap.
  3. 3Call `tiktok_comments_from_post` with `post_url` set to the three competitor videos whose captions describe the product, `max_results` 100 (0.4 credits each) to capture the complaints that repeat underneath.
  4. 4Call `b2b_ads_search` with `company_domain` set to the competitor's domain (0.5 credits) to check whether they are paying to amplify the message that ranks organically on TikTok. Use `b2b_ads_search` here, not `meta_ads_analysis` — that one needs a Facebook Page ID or page URL, not a domain.

Outcome: A one-page Monday snapshot for 2.5 credits: video and account counts for you versus the competitor, the three complaints that repeat under their content, and whether that content is paid-amplified.

Prompts

Paste any of these into Claude once the skill is installed and the SyncGTM MCP is connected.

Search TikTok for "cold email tips" and return 50 results. Group them by author and show me every account that appears more than once.
Search TikTok for #b2bsales with max_results 50, then list the 10 accounts posting most in the results and what each one covers.
Search TikTok for "sales automation" and return 50 results. Tell me which angles repeat in the captions and which angle nobody is covering.
Run tiktok_search_query on "outbound agency" with max_results 25, then pull the recent feed for the top 3 handles with tiktok_profile_video at max_posts 50.
Search TikTok for "AI SDR" with 50 results, pick the 5 videos whose captions name a tool, and pull their comments with tiktok_comments_from_post at max_results 100.

Best practices

  • Run the hashtag form and the plain keyword form as two separate calls. #coldemail and "cold email" surface different videos, and merging them roughly doubles the account pool for 0.8 credits.

  • Always set `max_results` to 50 or 100 for research. Cost is flat per call, so a limit of 10 saves nothing and hides the repeat-author pattern that makes the result useful.

  • Keep one topic per `query`. Two topics in one string dilutes both result sets — split them into two calls instead. There is no date, region or sort filter, so narrow by wording or filter the returned set yourself.

  • Never rank results by engagement. Each result carries only `description`, `author` and `url` — no view, like or follower counts — so rank accounts by how often they repeat.

  • Switch to `tiktok_profile_video` (0.5 credits, `username` or `profile_url` plus `max_posts`) the moment you know the handle, and to `tiktok_comments_from_post` (0.4 credits per video, `post_url` plus `max_results`) when you want what the audience said rather than what the creator posted.

  • Results are a snapshot of what TikTok surfaces at call time. Timestamp every output and re-run the same query on a schedule before you call anything a trend.

Complementary skills

Explore more

Ready to run /tiktok-search-query on your own data?