Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed LinkDigest client that sends user-provided public media links to linkdigest.dev for transcript and summary results.
Use this skill only for links you are comfortable sending to linkdigest.dev. Do not submit private or access-controlled content, keep LINKDIGEST_API_KEY in environment/config rather than chat or source files, and prefer the bundled Python client over the optional unpinned npx MCP bridge unless you pin and trust that package.
The skill invokes shell commands, reads an environment variable, and performs outbound network requests, but it does not declare any explicit tool scope or permission boundaries. In agent environments, this weakens least-privilege controls and can allow the skill to run with broader capabilities than users or orchestrators expect.
The manifest description is entirely in Chinese and describes when the skill should be used, but it does not indicate that language is optional or that the skill is intended only for a Chinese-speaking or region-specific context. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -sS -X POST https://linkdigest.dev/api/v1/digest \
-H "Authorization: Bearer $LINKDIGEST_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url": "https://v.douyin.com/<短链>/", "format": "json"}'
The documentation tells users to run npx mcp-remote without pinning a version, which causes execution of whatever package version is current at install time. If the package is compromised or a breaking update is published, users may run unreviewed code with their bearer token and network access.
The manifest presents this skill as focused on long-video transcription with timestamps, frame OCR, chapters, and notes for long videos. However, the CLI description and options show a broader link-digestion tool that reads Xiaohongshu/Douyin/TikTok/YouTube/X/WeChat article links into text, supports generic digest modes, translation, and 'breakdown' analysis, indicating substantially wider behavior than the manifest claims.
The manifest describes producing transcripts, frame text, chapters, notes, and summaries from long videos. The code additionally exposes '--breakdown' for '爆款拆解' and '--translate-to' for translated output, which are separate content-transformation features not claimed in the manifest description.
该文件的模块说明、用法和后续大量用户可见日志/错误信息均固定为中文，属于自然语言层面的默认语言强制。规则要求在未提供语言/locale 选择、opt-in 或明确合理地域限制时，语言强制应被视为政策问题。
The module docstring explicitly states '只连 linkdigest.dev 这一个域名'. Later, the error hint advises users they can '改用 SKILL.md 里的 curl 写法', which implies the documented operational path may involve behavior outside this script's claimed exclusive network scope. This creates intent ambiguity in the documentation even though the code here itself targets linkdigest.dev.
No suspicious patterns detected.