Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Doubao Video Creator
v2.0.0基于火山引擎豆包视频生成模型，辅助用户配置项目并分场景确认提示词，逐步生成并合成专业短视频。
⭐ 0· 66·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The skill's name, description, and code (prompt_generator, video_project, doubao_video_creator, etc.) are consistent with a Doubao-based video creation assistant. It legitimately asks for a Doubao API key and offers project-level config. However, many publishing-related files and examples include unrelated credentials (GitHub PAT, ClawHub token) and push/publish scripts that are not required for runtime video generation, which is unexpected and increases risk.
Instruction Scope
SKILL.md instructs the agent to collect and persist API keys to ~/.openclaw/workspace/doubao-config.json and project.json — reasonable for a project tool — but the runtime docs also reference explicit, hard-coded example keys and other external API keys (通义万相) and an absolute /root path (/root/.openclaw/workspace/wanxiang_generate.py). The instructions and included scripts reference publishing workflows and commands that accept tokens; the presence of hardcoded credentials and instructions to run publish/push scripts broaden the scope beyond the stated creation flow and risk accidental credential exposure or exfiltration.
Install Mechanism
There is no explicit install spec (instruction-only in registry metadata) which reduces installer risk, but the skill package contains many code files and shell scripts (publish.sh, quick_push_github.sh). No external downloads are specified. The build/install approach itself is not high-risk, but included scripts that perform git/publish operations and reference credentials raise concern if executed.
Credentials
Registry metadata declares no required env vars, but the skill expects the user to provide Doubao and optionally an image-generation API key. More problematic: multiple files contain hardcoded example secrets (Doubao example key, a 通义万相 API key, a GitHub PAT, and a ClawHub token). Publishing logs and guides include tokens and example credentials which are unrelated to normal skill runtime and are disproportionate to the task of generating videos.
Persistence & Privilege
The skill persists API configuration and project state to ~/.openclaw/workspace and per-project project.json (expected for project workflows). That by itself is OK. Concern arises because scripts and docs encourage committing/pushing repos and include hardcoded tokens — a skill with writeable local config plus embedded publish scripts could be used (accidentally or maliciously) to exfiltrate credentials or push data if those scripts are executed.
Scan Findings in Context
[hardcoded_doubao_api_key_in_docs] expected: SKILL.md and examples include a Doubao API Key example string. An example value for the service the skill integrates with is plausible, but hardcoded keys in distributed files are risky — they may be real or encourage unsafe copy/paste.
[hardcoded_wanxiang_api_key_in_docs] expected: The SKILL.md and examples include a 通义万相 (image-generation) API key. If the skill supports image-generation integration this is relevant, but embedding keys in files is inappropriate and risky.
[hardcoded_github_pat_in_publish_guide] unexpected: FINAL_PUBLISH_GUIDE.md and related docs contain a GitHub personal access token (ghp_...) and username/password example used for pushing to GitHub. This is unrelated to video generation and should never be present in published skill files.
[hardcoded_clawhub_token_in_publish_log] unexpected: CLAWHUB_PUBLISH_LOG.md contains a ClawHub auth token (clh_...). This token is unrelated to runtime functionality and its presence indicates sensitive information leaked into the package.
What to consider before installing
This skill largely does what it says (help create Doubao videos) but contains red flags you should address before installing or using it:
- Do NOT assume any example API keys/tokens in the files are safe to use — treat them as leaked secrets. If you or your organization used any of those values, rotate the tokens immediately.
- Inspect the code (especially publish.sh, quick_push_github.sh, FINAL_PUBLISH_GUIDE.md, CLAWHUB_PUBLISH_LOG.md and any scripts) for hardcoded credentials or commands that push data. Do not run publish/quick-push scripts without reviewing them.
- Prefer to run the skill in a restricted/sandboxed environment and avoid running included shell scripts automatically. Verify that the skill asks you interactively for your Doubao API key rather than pulling any environment variables you haven't provided.
- Search the package for patterns like 'ghp_', 'clh_', 'API_KEY', 'sk-' and remove or redact any embedded secrets before use or publication.
- If you plan to install from a remote source, prefer a verified upstream repository (official GitHub URL) and confirm the repo owner/trustworthiness. If uncertain, ask the author for a cleaned release without embedded secrets.
If you want, I can: (1) list exact files/lines that contain credential-looking values, (2) suggest safe config handling code snippets (prompted input + secure storage), or (3) produce a short checklist to sanitize the repo before installing.Like a lobster shell, security has layers — review code before you run it.
latestvk979ysxbd3vmdq2bqthc3ktn3x842pqf
License
MIT-0
Free to use, modify, and redistribute. No attribution required.