Tiktok API

TikTok API — Transport + Ingestion for a Real Account

You own the wire: authenticate to a TikTok account, publish video, pull the numbers, and write those numbers into the wiki as a durable feedback log. You do not decide what to make, when to post it, or how to caption it — that is the shortform strategy/packaging family. Deliver clean transport and a queryable log; let the siblings interpret.

TikTok splits across three separate APIs, and a real account touches all three:

• Content Posting API — https://open.tiktokapis.com/v2/post/publish/... — the write side: init a publish, transfer the file, poll status. Audit-gated.
• Display API — https://open.tiktokapis.com/v2/video/... — the cheap read side: your own profile and basic per-video counters (view_count, like_count, comment_count, share_count).
• TikTok API for Business — the rich read side: watch time, completion, impression sources. Enabled through a separate business portal, not the standard developer app.

Auth is user OAuth v2 via Login Kit, never a service token. A human owns the account; you act on their behalf with a refresh token. There is no official TikTok SDK — you call the REST endpoints directly with any HTTP client. Treat the access token as a short-lived, refreshable credential object, never a hardcoded literal.

When to use / When NOT

Use when:

• Wiring a script or agent to publish to an account: Direct Post or upload-to-draft via /v2/post/publish/video/init/ (or /inbox/ for a draft), then FILE_UPLOAD chunked PUT or PULL_FROM_URL, then poll /v2/post/publish/status/fetch/.
• Pulling an account's own video stats: counters via Display POST /v2/video/query/ (or /v2/video/list/); watch-time / completion / impression-source via the Business Account API.
• Building the recurring "fetch performance → write to 02-DOCS/wiki/shortform/" loop that turns API responses into an account feedback log siblings can read.
• Debugging TikTok-specific failures: scope_not_authorized, url_ownership_unverified, rate_limit_exceeded (6 req/min), 24-hour access-token expiry, audit/video.publish not approved, unaudited-app private-only posting.

Do NOT use when (route to the sibling that owns it):

One line: this skill authenticates, calls, and ingests TikTok's Content Posting + Display + Business APIs into the wiki. What to post and how to package it belong to the shortform-strategy / shortform-ideation / shortform-packaging siblings; multi-network posting belongs to social-publisher.

1. One-time setup (do this before any code)

A checklist, because each missing step produces a distinct, confusing failure later:

1. Register a TikTok developer app in the developer portal.
2. Add the products you need: Login Kit (OAuth), Content Posting API (publish), Display API (read counts). Insights live in the separate TikTok for Business portal — enable that account access too if you need watch time/completion.
3. Set an exact redirect URI for the OAuth flow.
4. Submit the app for audit before posting public content. An unaudited app can only post privately (SELF_ONLY) and only to a limited set of test users. This is the #1 "works on my machine, breaks in prod" surprise — see rule below.
5. If you publish by URL (PULL_FROM_URL), verify the domain / URL-prefix in the portal (DNS TXT or URL-prefix), or every init returns url_ownership_unverified.

The three gates are independent. Do not assume one approval covers everything:

Bad:  "My app is approved, so publish + insights both work."
Good: Content Posting *audit* gates public publish;
      Display *scope* (video.list) gates own-video counts;
      Business *portal* access gates watch time / completion / impression sources.
      Three separate gates — check each.

Scope table — request only what the job needs:

Full app-registration + product-enable walkthrough, the audit gate, and scope_not_authorized troubleshooting live in references/oauth-setup.md.

2. Get an authed client and keep the token alive

OAuth v2: send the user to https://www.tiktok.com/v2/auth/authorize/, receive a code at your redirect URI, exchange it at https://open.tiktokapis.com/v2/oauth/token/, and store the refresh token.

The lifecycle is the load-bearing fact: access token expires in 24 hours (expires_in: 86400); refresh token lasts 365 days (refresh_expires_in: 31536000) and renews without user re-consent. So a daily-pull cron MUST refresh the access token every run, and a long-idle account silently dies at the 365-day refresh boundary.

# python: raw REST, no official TikTok SDK. requests/httpx both fine.
import time, json, os, requests

TOKEN_URL = "https://open.tiktokapis.com/v2/oauth/token/"
STORE = "tiktok_token.json"   # gitignored — holds the rotating refresh_token

def load(): return json.load(open(STORE)) if os.path.exists(STORE) else {}
def save(t): t["obtained_at"] = int(time.time()); json.dump(t, open(STORE, "w"))

def access_token():
    t = load()
    fresh = t.get("access_token") and time.time() < t.get("obtained_at", 0) + t["expires_in"] - 60
    if fresh:
        return t["access_token"]
    r = requests.post(TOKEN_URL, data={                 # refresh every run, 24h expiry
        "client_key": os.environ["TIKTOK_CLIENT_KEY"],
        "client_secret": os.environ["TIKTOK_CLIENT_SECRET"],
        "grant_type": "refresh_token",
        "refresh_token": t["refresh_token"],            # 365-day lifetime; rotates
    }, headers={"Content-Type": "application/x-www-form-urlencoded"})
    r.raise_for_status()
    new = r.json()
    save(new)                                           # persist the NEW refresh_token
    return new["access_token"]

// node: built-in fetch, no SDK.
import fs from "node:fs";
const STORE = "tiktok_token.json";

async function accessToken() {
  const t = JSON.parse(fs.readFileSync(STORE, "utf8"));
  if (t.access_token && Date.now() / 1000 < t.obtained_at + t.expires_in - 60) return t.access_token;
  const r = await fetch("https://open.tiktokapis.com/v2/oauth/token/", {
    method: "POST",
    headers: { "Content-Type": "application/x-www-form-urlencoded" },
    body: new URLSearchParams({
      client_key: process.env.TIKTOK_CLIENT_KEY,
      client_secret: process.env.TIKTOK_CLIENT_SECRET,
      grant_type: "refresh_token",
      refresh_token: t.refresh_token,
    }),
  });
  const n = await r.json();
  n.obtained_at = Math.floor(Date.now() / 1000);
  fs.writeFileSync(STORE, JSON.stringify(n));          // persist rotated refresh_token
  return n.access_token;
}

Rule: persist the refresh token and re-read it each run, never a bare access_token. A hardcoded access_token=... literal is a guaranteed failure within 24 hours — and is exactly what verify.sh flags.

Full token-exchange flow (authorize URL params, PKCE, code exchange) is in references/oauth-setup.md.

3. Publish a video (init → transfer → poll)

Publishing is always three steps: init the publish, transfer the bytes, poll until processing finishes (it is async). Pick the transfer mode first:

Init (FILE_UPLOAD) — returns publish_id and an upload_url:

import math, requests

CHUNK = 10 * 1024 * 1024                      # 10 MB, inside the 5–64 MB window
size = os.path.getsize("clip.mp4")
chunk_count = 1 if size < 5 * 1024 * 1024 else math.ceil(size / CHUNK)

init = requests.post(
    "https://open.tiktokapis.com/v2/post/publish/video/init/",
    headers={"Authorization": f"Bearer {access_token()}",
             "Content-Type": "application/json; charset=UTF-8"},
    json={
        "post_info": {"title": "caption #fyp", "privacy_level": "SELF_ONLY"},  # public needs audit
        "source_info": {
            "source": "FILE_UPLOAD",
            "video_size": size,
            "chunk_size": CHUNK if size >= 5 * 1024 * 1024 else size,
            "total_chunk_count": chunk_count,
        },
    }).json()
publish_id = init["data"]["publish_id"]
upload_url = init["data"]["upload_url"]

Transfer — PUT chunks sequentially to upload_url with a Content-Range header. Chunk min 5 MB, max 64 MB (final chunk up to 128 MB), 1–1000 chunks; a file under 5 MB is one chunk equal to the file size. Each PUT returns 206 (more to send) or 201 (last chunk accepted):

with open("clip.mp4", "rb") as f:
    for i in range(chunk_count):
        first = i * CHUNK
        data = f.read(CHUNK)
        last = first + len(data) - 1
        r = requests.put(upload_url, data=data, headers={
            "Content-Type": "video/mp4",
            "Content-Range": f"bytes {first}-{last}/{size}",  # exact byte span
        })
        assert r.status_code in (206, 201), r.text     # 206 = continue, 201 = done

Poll — TikTok processes asynchronously; check status until PUBLISH_COMPLETE. Respect the cap below — do not tight-loop:

import time
while True:
    s = requests.post(
        "https://open.tiktokapis.com/v2/post/publish/status/fetch/",
        headers={"Authorization": f"Bearer {access_token()}",
                 "Content-Type": "application/json; charset=UTF-8"},
        json={"publish_id": publish_id}).json()
    status = s["data"]["status"]
    if status in ("PUBLISH_COMPLETE", "FAILED"):
        break
    time.sleep(10)                                      # 6/min cap — sleep, never spin

Rate limit: 6 requests/minute per user access token → rate_limit_exceeded. Throttle init/status calls and back off; a tight status-poll loop blows the budget in seconds.

PULL_FROM_URL requires the domain/URL-prefix to be verified in the portal (HTTPS only, no redirects, 1-hour download timeout) or init returns url_ownership_unverified. Full PULL_FROM_URL init body and verification steps are in references/metrics-and-publish.md.

4. Pull performance (two APIs, one rule)

The load-bearing distinction: Display gives you counters; only the Business API gives you watch time, completion, and traffic.

# (a) Display API — basic counters only. scope video.list, up to 20 ids/request.
counts = requests.post(
    "https://open.tiktokapis.com/v2/video/query/",
    params={"fields": "id,title,view_count,like_count,comment_count,share_count,duration,create_time"},
    headers={"Authorization": f"Bearer {access_token()}",
             "Content-Type": "application/json"},
    json={"filters": {"video_ids": ["<id1>", "<id2>"]}}).json()
# returns: view_count, like_count, comment_count, share_count, duration, title, create_time

# (b) Business Account API — the real engagement signal.
# Returns the metrics Display CANNOT: average_time_watched, total_time_watched,
# full_video_watched_rate (completion), impression_sources (FYP / Following / profile /
# search), audience_countries. (Endpoint shape in references/metrics-and-publish.md.)

Bad:  expect average_time_watched / full_video_watched_rate from /v2/video/query/
Good: counters from Display /v2/video/query/;
      watch time + completion + impression_sources from the Business Account API.

Caveat: Business insight metrics lag 24–48h and can differ from the in-app numbers. Treat a fresh pull as provisional — the wiki log (next section) is where you watch them settle. Full metric catalog split by API is in references/metrics-and-publish.md.

5. Ingest into the wiki — the actual deliverable

A pull that prints to stdout and vanishes is wasted. Every pull appends a dated entry under 02-DOCS/wiki/shortform/, platform-namespaced, so the account's numbers become queryable history the strategy/packaging siblings can read.

02-DOCS/wiki/shortform/
  index.md                       # rolling pointer to latest snapshot + open questions
  tiktok-account-2026-06-02.md   # dated account snapshot (one per pull)
  videos/tiktok-<video_id>.md    # per-video running log, newest entry on top

Filenames carry the tiktok- prefix because the same shortform/ wiki may also hold Instagram and YouTube pulls — namespacing keeps platforms from colliding.

Per-pull entry template. The frontmatter is OKF v0.1 conformant — a non-empty type is the only hard requirement; title/tags/timestamp are the recommended OKF surface; the domain date/range/account/platform/source keys the siblings parse are preserved additively (OKF tolerates extra keys). date is the reporting day; timestamp is the ISO 8601 write moment:

---
type: shortform-metrics
title: TikTok account snapshot — 2026-06-02
tags: [tiktok, metrics, snapshot]
timestamp: 2026-06-02T09:00:00Z
date: 2026-06-02
range: 2026-05-26..2026-06-01
account: <open_id>
platform: tiktok
source: display-api + business-account-api
---
## KPIs
views: 52,140 | likes: 3,902 | comments: 211 | shares: 488

## Watch
full_video_watched_rate: 28.4% | avg_time_watched: 6.1s | total_time_watched: 88h

## Impression sources (top 3)
For You 71% · Personal profile 14% · Search 7%

## What changed since last pull
completion +3.1pts after the tighter cold-open; FYP share up 5pts.

Rule: append, never overwrite. The feedback log is the value — overwriting yesterday's snapshot destroys the trend the siblings need, and erases the 24–48h settling you only see across pulls. The per-video log (videos/tiktok-<id>.md) is an OKF append-log: write its frontmatter header once, prepend each new dated block newest-first, never edit a past block. index.md is the OKF reserved file — no frontmatter, standard markdown links only. Exact file tree, frontmatter, naming, and how siblings read the log: references/wiki-schema.md.

6. Rate & failure math

The publish token is capped at 6 requests/minute. Wrap publish/status calls in a token-bucket or backoff-with-jitter helper, and refresh the access token (24h expiry) before each cron run.

Error → cause map:

Anti-patterns

Cross-references

• ../social-publisher/SKILL.md — when the asset goes to many networks, not just TikTok.
• ../instagram-api/SKILL.md — same family pattern, Instagram's Graph/Content Publishing API.
• ../youtube-api/SKILL.md — same transport+ingestion shape, YouTube's two APIs.
• ../remotion-video/SKILL.md — produce the clip file this skill only uploads.
• shortform-strategy, shortform-ideation, shortform-packaging, shortform-editing (catalog ids) — what the numbers mean, what to make, and how to package/edit it.
• api-connector-builder, automation-flows, knowledge-ops (catalog ids) — generic connector wrapping, cross-tool chaining, and wiki conventions.