tiktok-create-master-skill
Skill15 starsChanged 36 days agoMITView on GitHub
Use for any TikTok Ads launch: a new campaign, ad group or ads, catalog and Smart+ campaigns. Creative-first, placement-aware, one-call launch via tiktok_propose_create_campaign_structure with approval before anything goes live.
In this repository61 other files it gives its agents.
- access-and-permissionsSkill
- accounts-and-connectionsSkill
- ad-brief-generatorSkill
- affiliate-marketingSkill
- analyse-creativeSkill
- automate-with-agentsSkill
- brief-creativeSkill
- build-reportsSkill
- clean-wasted-spendSkill
- competitor-ad-intelligenceSkill
- creative-psychology-hooksSkill
- creative-researchSkill
- diagnose-performanceSkill
- email-marketingSkill
- ga4-source-of-truthSkill
- get-startedSkill
- gomarble-skills-creative-strategistsSkill
- gomarble-skills-ecommerce-brandsSkill
- gomarble-skills-google-ads-auditSkill
- gomarble-skills-meta-ads-auditSkill
- gomarble-skills-saas-companiesSkill
- google-ads-create-master-skillSkill
- google-ads-depth-of-analysisSkill
- google-ads-guardrailsSkill
- google-ads-keywordplannerSkill
- google-ads-pmax-evaluationSkill
- google-ads-pmax-scalingSkill
- google-ads-search-analysisSkill
- google-ads-search-executionSkill
- google-ads-shoppingSkill
- google-ads-tool-fundamentalsSkill
- google-impression-shareSkill
- google-pmax-pulseSkill
- google-search-auditSkill
- google-search-term-auditSkill
- launch-campaignsSkill
- linkedin-adsSkill
- manage-campaignsSkill
- meta-ads-auditSkill
- meta-agent-operationsSkill
- meta-create-master-skillSkill
- meta-creative-analysisSkill
- meta-creative-fatigue-detectionSkill
- meta-creative-strategySkill
- meta-custom-event-interpretationSkill
- meta-daily-optimizationSkill
- meta-depth-of-analysisSkill
- meta-guardrailsSkill
- meta-performance-analysisSkill
- meta-tool-fundamentalsSkill
- microsoft-adsSkill
- organic-socialSkill
- own-creative-diagnosisSkill
- research-competitorsSkill
- search-console-master-skillSkill
- shift-budgetSkill
- shopify-order-disciplineSkill
- tiktok-adsSkill
- winning-ads-engineSkill
- winning-ads-orchestratorSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 39
Never keep varying one parameter to escape an error: If a create fails on targeting or placement, the fix is almost always a *different* …
- Distinctive: line 71
Exception — catalog / shop ads: If the user asks to promote a *product feed or store* rather than a specific creative ("run catalog ads", …
- Distinctive: line 112
If the user supplied a still image and wants US or UK delivery, say so plainly before proposing: image ads cannot serve there on most …
- Very specific: line 132
Available locations for the chosen placement: Call /tool/region/ with placements + objective_type (tiktok_search_targeting surfaces this).
- Distinctive: line 134
Never assume a country is eligible, and never copy a country list from documentation — the live list differs per account and changes.
- Distinctive: line 144
Default to proposing. Ask only when a choice genuinely changes the outcome: multiple ad accounts, an ambiguous funnel stage, or no …
- Distinctive: line 178
Never split into one call per entity.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
TikTok Ad Launch Workflow
A creative-first workflow for launching TikTok ads. The sequence is fixed: creative → placement → account scan → propose. Placement is decided second, immediately after the creative, because on TikTok the creative format dictates the placement and placement can never be changed after the ad group exists.
Tool behavior: the entire launch — campaign, ad group and ads — is proposed in ONE call to
tiktok_propose_create_campaign_structure. Each entity gets its ownoperation_id. Through AI app connections TikTok changes can only be validated as a dry run: show the user the validated launch, then give them the steps to apply it in TikTok Ads Manager (seetiktok-ads). Where a live call is possible, it executes in dependency order (campaign → ad group → ads). Never thread IDs between slots — the server wires each new parent's ID into its children.
HARD RULE — Creation Failure Limit (strictly enforced)
If a launch call fails 2 or more times within the same workflow, STOP. Do not propose again, do not retry, do not attempt a workaround.
"I've attempted this step twice and it wasn't successful both times. This looks like it needs a change in TikTok Ads Manager or an account permission I can't resolve from here. I'd suggest reviewing the error above before trying again in a new conversation. Is there anything else I can help you with?"
A "failure" is a validation error (601), a platform error (403), or a thrown
exception. A user rejecting a card is NOT a failure. Partial success is NOT a
failure — re-propose only the failed children with adgroup_id.
Never keep varying one parameter to escape an error. If a create fails on targeting or placement, the fix is almost always a different field than the one you just changed. Guessing countries one at a time is the classic wrong loop.
Sub-Skill Index
| Slot | Read before filling | Path |
campaign object | Building the campaign slot | references/campaign.md |
adgroup object | Building the ad group slot | references/ad-group.md |
ads array | Building the ads | references/ad.md |
| Upgraded Smart+ (any smart_plus work) | Before touching tiktok_propose_manage_smart_plus | references/smart-plus.md |
Slot combinations (each parent is a config to create XOR an existing ID):
campaign+adgroup+ads→ full launchcampaign_id+adgroup+ads→ new ad group + ads under an existing campaignadgroup_id+ads→ add ads to an existing ad group- ONE ad group per call. More ad groups → one call each, passing
campaign_id.
Phase 1 — Get the Creative
The creative comes first, always.
- Image, video or carousel attached → go to Phase 2.
- Referenced an existing ad → pull it and go to Phase 2.
- Nothing attached → ask exactly one question: "Share the creative, or point me to an existing ad to clone." No objective, budget or audience questions.
Exception — catalog / shop ads. If the user asks to promote a product feed or store rather than a specific creative ("run catalog ads", "promote my products", "GMV Max"), there is no creative to collect. Skip to the inventory check in Phase 3 instead, and pick the path from what actually exists:
| Inventory found | Path |
A catalog (resource: CATALOG) | PRODUCT_SALES + campaign_product_source: CATALOG → catalog ad group + ads |
Only a TikTok Shop store (resource: STORE) | GMV_MAX — campaign-only, no ad group or ads |
| Neither | Stop and say so. No tool here creates a catalog; it must be set up in TikTok Ads Manager / Business Center first. |
Having a store does not imply having a catalog — most Shop accounts have a store and zero catalogs. Check before promising either path.
Phase 2 — Decide the Placement (TikTok-specific, do this early)
The creative format decides the placement, and the placement decides everything downstream. Get this wrong and the ad group must be rebuilt — it cannot be edited.
| Creative | Placement | Identity type |
| Video | PLACEMENT_TIKTOK or automatic | BC_AUTH_TT / TT_USER |
| Carousel | PLACEMENT_TIKTOK | BC_AUTH_TT / TT_USER |
| Single image | PLACEMENT_PANGLE or PLACEMENT_GLOBAL_APP_BUNDLE only | CUSTOMIZED_USER |
Three rules that cause most launch failures:
- A single image ad cannot run on TikTok feed. Doc 1777633230937090 — image
ads exist only on Pangle and Global App Bundle. On TikTok placement the create
fails with
Incorrect source field (40002), which names nothing. - The identity type must match the placement. Spark identities
(
BC_AUTH_TT,TT_USER) exist only on TikTok. Pangle and Global App Bundle take the advertiser's ownCUSTOMIZED_USERidentity. Mixing them gives the sameIncorrect source field. - Pangle and Global App Bundle serve far fewer countries than TikTok, and the list depends on the ad account's registration market. Verify before proposing — see Phase 3.
If the user supplied a still image and wants US or UK delivery, say so plainly before proposing: image ads cannot serve there on most accounts, and the options are a Pangle-eligible country or a video creative.
Phase 3 — Scan the Account
Pull what you need to propose without asking. Synthesise into 1–2 lines; never dump raw data.
- Existing campaigns/ad groups — objective, optimization goal, budgets, geos
tiktok_get_identities— note which types exist. You need aCUSTOMIZED_USERidentity for image ads and aBC_AUTH_TT/TT_USERone for video. If the required type is missing, that's a blocker worth stating early.- Pixel(s) via
tiktok_list_pixelswhen optimizing for CONVERT/VALUE - Catalog / store inventory via
tiktok_list_catalogs— only for the catalog or GMV Max paths.resource: CATALOG→PRODUCT_SET→PRODUCT(individual SKUs), orresource: STOREfor TikTok Shop. This is a read-only listing; it cannot create a catalog, a product set, or a store. - Available locations for the chosen placement. Call
/tool/region/withplacements+objective_type(tiktok_search_targetingsurfaces this). Never assume a country is eligible, and never copy a country list from documentation — the live list differs per account and changes.
Synthesis output: "Ref: <campaign> — Website conversions / CONVERT+SHOPPING
/ US 18-55 / $25/day. Pixel <n>, identity <name> (BC_AUTH_TT)."
Phase 4 — Propose or Ask
Default to proposing. Ask only when a choice genuinely changes the outcome: multiple ad accounts, an ambiguous funnel stage, or no reference data and unknown budget. Batch into one turn, max 2 questions, tappable options.
Phase 5 — Show the Configuration
| Field | Proposed value | Source |
| Objective | Website conversions | Mirrors top campaign |
| Placement | Pangle (image creative) | Creative format |
| Locations | Canada | Pangle-eligible, from /tool/region/ |
| Optimization | CONVERT / SHOPPING | Account default |
| Daily budget | $25 | Median of existing ad groups |
| Identity | EvanAlexanderGrooming (CUSTOMIZED_USER) | Required for Pangle |
| Status | Paused | Always |
Then send the proposal as a dry run and show the user exactly what it would create.
Phase 6 — Launch (ONE call)
tiktok_propose_create_campaign_structure({
"advertiser_id": "...",
"currency_code": "USD",
"campaign": { ...per references/campaign.md... },
"adgroup": { ...per references/ad-group.md... },
"ads": [ ...per references/ad.md... ]
})- Never split into one call per entity. Never pass IDs between slots.
- Everything is created PAUSED (
operation_status: DISABLE). TikTok's own default is ENABLE — i.e. live spend on approval — so the server forces DISABLE unless the user explicitly asks otherwise. Enable after review. - The dry run persists operations; the live call takes only
operation_ids+account_id. Do not resend the full payload on the live call.
Other campaign families — dedicated tools, not this launch flow
- Upgraded Smart+: reads
tiktok_get_smart_plus, writestiktok_propose_manage_smart_plus— loadreferences/smart-plus.mdfirst; the payload rules differ from everything in this skill. - Reach & Frequency (reservation buying): estimate with
tiktok_get_reach_frequency, thentiktok_propose_manage_reach_frequency. CREATE_ADGROUP RESERVES REAL BUDGET on approval and cannot be paused; the account must be R&F-allowlisted AND hold a Commercial Contract for Branding. - Split tests:
tiktok_get_split_test/tiktok_propose_manage_split_test— CREATE overwrites both ad groups' budgets and requires them ENABLED. - Appeals & offline events:
tiktok_get_offline_event_sets/tiktok_propose_manage_review_and_events— an appeal is one-shot per rejection.
Reading TikTok's errors
TikTok returns 40002 for almost everything and rarely names the field. The
server rewrites the two worst offenders — trust its added explanation.
| Message | What it actually means |
Incorrect source field | The identity type doesn't match the placement (or an image ad is on TikTok placement). Not a creative problem. |
Unable to access image… | You passed material_id where image_ids wants the image_id (ad-site-i18n-sg/...). |
You must upload an image | A video ad has no cover — pass image_ids with the cover. |
audience … must increase to over 1000 | The targeted countries have no eligible inventory on the chosen placement. Change the placement or the countries — not the audience size. |
Custom identities are no longer supported | CUSTOMIZED_USER on TikTok placement. Use BC_AUTH_TT/TT_USER there. |
shopping_ads_type rejected as an invalid enum | You used the campaign-level values (PRODUCT/LIVE) on the ad group, or the ad-group values (VIDEO/LIVE/PRODUCT_SHOPPING_ADS) on a GMV_MAX campaign. Same field name, different enum per slot. |
Copy API is an allowlist-only feature | Campaign copy (regular AND Smart+) is not granted to this advertiser. Stop — a TikTok representative must enable it. Retrying cannot help. |
CTA not supported repeated across different values | Account-enablement wall (Smart+ ads), not a payload problem. Stop instead of iterating CTA values. |
Split tests can only be created for active ad groups | The split-test endpoint requires ENABLED ad groups at creation time, even with a future start_time. |
description: Missing data for required field (offline event sets) | /offline/create/ requires description even though the doc table doesn't mark it. |
Your budget setting must not be less than $50 (or similar) | Some accounts enforce a HIGHER floor than the documented $20 baseline per objective. Re-propose at the number TikTok names. |
Never re-upload a creative in response to any of these. The asset is fine; the configuration is not.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

