Decision intelligence for TikTok Shop

Read the signal.Make the move.

An open-source intelligence framework that turns TikTok Shop signals into hypotheses, experiments and operating decisions.

$tiktok-intelligence Run today's brief.

v3.1.0 · One HTML file · No setup questionnaire · MIT License

24–72hDefault research window
≤ 3Operating actions
0Required setup fields
MITOpen-source license

The operating method

News is input.Decisions are output.

The Skill verifies what changed, connects it to business economics, then proposes the smallest experiment that could change a decision.

01 / SIGNAL

Signal

Detect a material platform, policy, ad, creator or market change.

02 / EVIDENCE

Evidence

Check source quality, event date, freshness and scope.

03 / IMPLICATION

Implication

Translate the change into operating and economic consequences.

04 / HYPOTHESIS

Hypothesis

State what the evidence strengthens, weakens or leaves unknown.

05 / EXPERIMENT

Experiment

Design a controlled test with metrics, thresholds and a stop rule.

06 / DECISION

Decision

End with TEST, SCALE, PAUSE, MONITOR, INVESTIGATE or IGNORE.

Before and after

Before: more updates

  • Repeated headlines without an event-date check
  • GMV and attributed ROAS treated as profit
  • One operator screenshot becomes a benchmark
  • Vague advice with no owner or threshold

After: better decisions

  • Freshness, source quality and scope verified
  • Contribution profit and missing costs made explicit
  • Operator claims remain labeled and bounded
  • Actions carry a decision label, metric and stop rule
Reasoning record / captured 2026-08-26 Public evidence only

Attributed GMV is not automatically incremental lift.

TikTok's official documentation says selected-product orders may be attributed across paid, organic and affiliate activity while a GMV Max campaign is active—even when the buyer did not engage with an ad.

Business implication01

Dashboard-attributed GMV alone cannot prove paid-media causality or contribution profit.

HypothesisH-ATTR-001

Budget allocation based only on attributed ROI can overstate the incremental economic value of paid spend.

ExperimentA/B

Use a matched control or holdout. Compare incremental contribution profit per ad-spend unit after one complete purchase cycle.

DecisionTEST

Scale only when control-adjusted contribution is positive and clears the business's predeclared meaningful-lift threshold.

What not to do Do not claim incrementality from attributed sales alone. Do not infer profit from GMV or attributed ROI without the missing cost inputs.

This is a public-source reasoning demonstration—not a private seller result, fabricated testimonial or claim of a new 72-hour platform change.

One engine, six lenses

It starts where you operate.

The report routes only the modules that matter to your operating model. A current instruction always overrides stored preferences.

LENS / 01

Seller or brand

Contribution profit, ad efficiency, creator economics and account risk.

LENS / 02

Agency

Client outcomes, retention, operating margin, activation and account risk.

LENS / 03

Creator network

Active creators, repeat posting, revenue per creator and retention.

LENS / 04

LIVE operator

Contribution per LIVE hour, traffic quality, host and compliance risk.

LENS / 05

SaaS or data

Merchant demand, product gaps and automation opportunities.

LENS / 06

Research or investment

Market structure, policy, competition and platform direction.

Three-step installation

From download to first brief.

The Release package contains the ready-to-install tiktok-intelligence folder. No questionnaire is required before the first run.

Download01

Get the official Release.

Download the versioned ZIP from GitHub Releases, where the asset download count is recorded without adding personal tracking.

Download ZIP ↓
Verify SHA-256 checksum

Compare with the latest Release notes ↗

macOS / Linux:
shasum -a 256 tiktok-intelligence.zip

Windows PowerShell:
(Get-FileHash .\tiktok-intelligence.zip -Algorithm SHA256).Hash.ToLowerInvariant()
Extract02

Place it in your Skills folder.

Extract the package so SKILL.md lives inside tiktok-intelligence/. Keep private config and state files outside public packages.

$skill-installer Install the skill from https://github.com/daisymyu-dev/TikTok-Commerce-Intelligence

Installer alternative. Release ZIP downloads are the installs visible in the GitHub asset count.

Manual ZIP extraction

macOS / Linux

mkdir -p ~/.agents/skills
unzip tiktok-intelligence.zip -d ~/.agents/skills

Windows PowerShell

New-Item -ItemType Directory -Force "$env:USERPROFILE\.agents\skills" | Out-Null
Expand-Archive .\tiktok-intelligence.zip -DestinationPath "$env:USERPROFILE\.agents\skills"

~/.agents/skills/tiktok-intelligence/SKILL.md

Upgrade safely: extract to a temporary folder first, preserve private config and state, then replace only public package files.

Run03

Refresh, then ask one sentence.

Refresh the host so it discovers the Skill. Codex uses $, while ChatGPT with Skills enabled uses @ and the Skill selector.

$tiktok-intelligence Analyze the last 72 hours of TikTok Shop developments.

ChatGPT: @tiktok-intelligence Give me today's TikTok Commerce Intelligence Brief.

Engine
SKILL.mdHow the system thinks—public and reusable.
Profile
config.local.yamlWho you are and what matters—private by default.
Memory
state/What the system learns—saved only with authorization.

Private by default

Public engine.Private context.

The public package contains generic methods, public sources and fictional schema examples. It excludes credentials and internal operating data.

  • No identity dataNo personal contacts, KYC or private account identifiers.
  • No internal economicsNo actual SKU, sales, margin, spend, supplier or team data.
  • No silent persistenceProfile or memory updates require user authorization.
  • No site trackingNo analytics, cookies, pixels or third-party runtime requests.

Default economic lens

Net sales − product cost − fulfillment − platform fees − advertising − affiliate commissions − samples − discounts − returns = contribution profit

GMV does not establish profit. When costs are missing, economic impact remains unconfirmed.

Install the decision loop

Stop collecting updates.Start changing decisions.

One intentional extra step

Downloading or installing does not automatically Star the repository. If the Skill helps you make a better decision, return to GitHub and add your Star intentionally.