Security audit

Human Distill

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed research skill for creating evidence-labeled persona notes from public web and Douyin content, with privacy precautions needed for logged-in browsing and saved dossiers.

Install only if you are comfortable letting the agent browse Douyin in a logged-in browser session and save collected public-person profile notes in your workspace. Use it for public content, avoid private or paid material, and delete generated memory/people files when you no longer need them.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Missing User Warnings

Medium

Confidence: 78% confidence
Finding: The README promotes browser-based scraping of Douyin content and saving extracted material into workspace files, but it does not clearly warn users about privacy, account-session exposure, retention of collected personal data, or the sensitivity of generated dossiers about individuals. In a skill specifically designed to profile people from public content, missing data-handling guidance can lead to overcollection, persistence of personal data, and unsafe downstream use of scraped content.

Vague Triggers

Medium

Confidence: 71% confidence
Finding: The skill relies on broad natural-language triggers such as “蒸馏”, “人物画像”, and related phrases, which can cause the skill to activate in contexts the user did not clearly intend, including profiling real people or initiating scraping-heavy workflows. Because the skill can use browser-based extraction and generate persistent persona files, overly broad activation increases the risk of unintended collection, profiling, and misuse.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill directs the agent to write multiple artifacts into `memory/` and `people/[slug]` as part of normal execution, but it does not require explicit user consent or a clear warning that workspace and gbrain files will be created or modified. This creates a real integrity/privacy risk because users may expect analysis-only behavior while the skill persistently stores scraped personal-profile data and derived summaries.

Missing User Warnings

High

Confidence: 97% confidence
Finding: The skill instructs use of a logged-in browser profile to access and scrape Douyin through the user's authenticated session, including checking login state and reusing persistent cookies, without a clear privacy/security warning about session exposure. This is dangerous because authenticated browsing can expose private account context, personalized data, or permit actions under the user's session if the browser tooling is overprivileged or the target site behavior changes.

Static analysis

No suspicious patterns detected.