Tiktok Ad Research: Install, Source and Security | FunnelSlayer

Tiktok Ad Research

Published by postplusai in postplus-skills

Review recommended499 installs

What this skill does

Research TikTok Creative Center or ad-library style datasets for winning ad patterns, regions, objectives, hook language, and creative signals without mixing paid ads with organic creator discovery.

Add Tiktok Ad Research to your agent

Review the source and files first. When you are ready, copy the prompt instruction or use the CLI command supported by your environment.

Install with a prompt

Paste this into a compatible coding agent:

add this skill "tiktok-ad-research" from https://github.com/postplusai/postplus-skills

Install with the CLI

Run this command in a controlled environment after reviewing the repository:

npx skills add https://github.com/postplusai/postplus-skills --skill tiktok-ad-research

Skill instructions

TikTok Ad Research

Use this skill for paid TikTok ad intelligence, not organic creator or content discovery.

Apply shared rulebook and user-guidance rules from postplus-shared.

Task Shapes

Use this skill when the user wants to:

  • find paid TikTok ad creatives that are performing well,
  • inspect top ads for a category, country, region, language, or objective,
  • study competitor paid hooks, selling points, and CTAs,
  • turn ad examples into ad briefs or creative implications.

Do not use this skill for creator discovery, community comments research, organic content lane mapping, or TikTok music workflows.

Collection Key Routing

Route paid-ad collection to the released hosted collection key for top paid ad examples and optional ad-level metrics. Discover the exact released collection key and request shape with postplus research schema --json, then pass the request body to postplus research collect <collectionKey> --request <input.json>.

Default Workflow

  1. Classify the request into a paid-ad task shape.
  2. Build a narrow first-pass request for the category, region, objective, or competitor scope.
  3. Write the request to a request file and run postplus research collect <collectionKey> --request <input.json>.
  4. Normalize the result into observed ad facts.
  5. Separate observed ad facts from inferred creative implications.
<!-- BEGIN GENERATED EXECUTION EXAMPLE -->
postplus research collect tiktok-ads-top --request request.json --output result.json
<!-- END GENERATED EXECUTION EXAMPLE -->

Keep raw datasets and intermediate files under .postplus/tiktok-ads/; keep final shortlists, summaries, or briefs where the user can inspect them.

Good Output

Return top brands or advertisers, dominant objectives, repeated hook and offer language, CTA patterns, region or language scope, duration distribution, and which metrics were actually present in the sample.

Failure Modes

  • Stop if the request is organic creator/content research; route to tiktok-research instead.
  • Stop on missing required input, unsupported key, missing auth, unavailable hosted service, or stable network failure.
  • Do not invent fallback execution paths, private backend calls, or web-search replacements for paid ad collection.

Handoff

  • Organic TikTok benchmark comparison -> tiktok-research.
  • Ad creative review after humans inspect outputs -> creative-qa.
  • Shot-level or spoken-line breakdown -> video-analysis.

Public Command Boundary

  • Choose the smallest matching command or workflow from the user input and run it directly.
  • Readiness diagnostics: postplus doctor --skill tiktok-ad-research.
  • If an owned CLI or script command fails, report the exact error and stop. Do not bypass the failure with metadata-only answers, readiness probing, local payload rewrites, fallback providers, or unpublished tools.
  • Use postplus research schema --collection-key <collectionKey> --json only when constructing or repairing an unknown request shape.
  • Hosted collection: postplus research collect <collectionKey> --request <input.json> --output <result.json> (input = the collection parameters).
  • Resume a pending collection: postplus research collect --run-handle <runHandle> --output <result.json>.
  • Keep the first pass bounded; expand only after inspecting the first result.
  • If the CLI returns a quote-confirmation challenge, run postplus quote confirm --json --challenge-file <challenge.json> and retry with the returned token.

Files included

  • SKILL.md

More skills