Douyin Influencer Outreach
WarnAudited by ClawScan on May 10, 2026.
Overview
The skill openly performs Douyin outreach, but it automates batch private messages from a browser profile and includes risk-control avoidance guidance, so it needs careful review before use.
Only install or run this if you are comfortable letting the agent use your Douyin browser session to message people. Before sending, insist on reviewing each recipient and message, understand the account-policy risk, and periodically review or delete the local contacted.csv file.
Findings (6)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent could send multiple private messages from the user's Douyin account, affecting account reputation and creating spam or policy-compliance risk.
After the user confirms parameters, the workflow tells the agent to automatically send DMs to every qualifying account, rather than requiring recipient-by-recipient approval.
找到符合条件的达人立即发送私信 ... ✅ 全部达标 → 立即发送私信 → 记录 CSV → 计数 +1
Require an explicit preview and approval for each recipient and message before sending, set hard rate limits, and make platform-policy risks clear.
A logged-in Douyin account could be used for outreach without the registry metadata clearly declaring that account/session authority.
The skill acts through the OpenClaw browser profile on Douyin and sends messages, which likely uses the user's logged-in session, but the metadata declares no primary credential or account boundary.
browser action=open profile=openclaw targetUrl="https://www.douyin.com/user/{sec_uid}" ... 找到并点击私信按钮Declare the Douyin account/session requirement, show which account is active before sending, and require user confirmation that this account should be used.
Message content or page-derived variables could break the generated script, and raw page evaluation increases the impact of mistakes in a logged-in Douyin session.
The helper builds JavaScript code by interpolating message text and then sends via a synthetic Enter event; this contradicts the SKILL.md claim to use real typing and can be fragile or unsafe if message variables contain special characters.
input.textContent = '${message.replace(/'/g, "\\'")}'; ... input.dispatchEvent(new KeyboardEvent('keydown', { key: 'Enter'Avoid raw page-evaluated message injection; use safer browser typing APIs, or at minimum serialize message strings with JSON.stringify and require a send preview.
The user may be encouraged to run outreach in a way that risks account restriction, platform-policy violations, or unwanted messages to recipients.
The examples include tactics framed around avoiding platform risk controls and account enforcement while conducting bulk outreach.
避免风控，间隔 30 秒 ... 避免敏感词（微信、电话、二维码、转账等） ... 新号建议先养号 1-2 周再开始拓展
Remove evasion-oriented guidance and replace it with clear compliance, consent, rate-limit, and opt-out practices.
A local record of contacted people and sent messages will remain on disk across runs.
The skill stores a persistent local contact history including profile details and message content for deduplication.
文件路径：`~/.openclaw/workspace/skills/douyin-influencer-outreach/contacted.csv` ... 昵称,抖音号,sec_uid,粉丝数,获赞数,IP属地,简介,私信内容
Tell users how to review, edit, and delete the CSV, and minimize stored fields to what is needed for deduplication.
If someone installs the package dependencies, they may get an unpinned version rather than a reviewed one.
Although the registry says there is no install spec, the package declares a wildcard dependency that could resolve to changing code if installed in another context.
"dependencies": { "openclaw": "*" }Remove unused dependencies for an instruction-only skill or pin dependency versions and provide a lockfile.