Security audit

Osint Personal Profiler

Security checks across malware telemetry and agentic risk

Overview

This skill is openly an OSINT profiler, but it gives agents a broad workflow for building detailed dossiers on arbitrary people from minimal identifiers.

Install only if you need an authorized privacy audit, defensive exposure assessment, or lawful due-diligence workflow. Do not use it to investigate private individuals without consent or a clear legal basis, and treat any generated profiles as sensitive local data that should be minimized, protected, and deleted when no longer needed.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (22)

Vague Triggers

Medium

Confidence: 83% confidence
Finding: The trigger list includes broad terms such as 公开信息收集, 人物背景调查, 数字足迹分析, and 开源情报收集, which can match ordinary privacy or research requests and invoke a highly invasive profiling workflow unintentionally. Because this skill performs sensitive cross-source profiling from minimal identifiers, accidental activation materially increases privacy and misuse risk.

Missing User Warnings

High

Confidence: 95% confidence
Finding: The description frames the skill as a system for reconstructing any person's full profile from public traces, but does not place a prominent up-front warning about privacy sensitivity, consent requirements, or data handling limits. For a skill centered on collecting and inferring personal attributes, relationships, and motives, this omission makes unsafe or unauthorized use substantially more likely.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill states that profile data is stored locally, but the description does not clearly warn users up front that collected personal profiles will be retained on disk or define retention/deletion controls. Silent or under-disclosed storage of sensitive OSINT-derived profiles increases the risk of secondary exposure, unauthorized reuse, or over-retention.

Missing User Warnings

High

Confidence: 98% confidence
Finding: The template operationalizes comprehensive profiling of a named individual across many sources and explicitly encourages cross-source inference of sensitive traits without any meaningful warning, minimization, or purpose limitation. Even if sources are nominally public, aggregating and inferring identity, family, health, relationships, motives, and behavioral patterns materially increases privacy harm and can enable stalking, doxxing, discrimination, or targeted social engineering.

Missing User Warnings

High

Confidence: 99% confidence
Finding: The skill explicitly operationalizes end-to-end collection and inference of a person's sensitive attributes from public data, but provides no meaningful consent, necessity, minimization, or anti-harm controls. Framing this as a reusable workflow for '任意个人' materially increases the risk of stalking, doxxing, targeted harassment, discrimination, and social-engineering preparation.

Missing User Warnings

High

Confidence: 99% confidence
Finding: These steps direct collection of family identity, estimated age, gender, body traits, nationality, ethnicity, language, and related personal characteristics without warning that these are highly sensitive and prone to misuse. Even when sourced from public content, aggregating and structuring them into a dossier creates new privacy and safety harms well beyond the original context of disclosure.

Missing User Warnings

High

Confidence: 98% confidence
Finding: The workflow goes beyond factual collection into relationship mapping and inference of hierarchy, motives, and life stage, which are classic profiling activities with substantial downstream decision risk. Without strong safeguards, these inferred judgments can enable manipulation, coercion, reputational harm, and unfair treatment while presenting uncertain conclusions as structured intelligence.

Missing User Warnings

High

Confidence: 97% confidence
Finding: This section operationalizes systematic discovery and correlation of a person's online accounts across many platforms using strong identifiers such as phone numbers, email addresses, exact URLs, username derivation, reverse-image search, and historical cache lookup, but provides no meaningful user-facing warning about privacy, consent, or abuse risk. Even though it references legality, the workflow materially increases a user's ability to identify, track, and profile a real person at scale, which can enable stalking, doxxing, targeted harassment, or social-engineering preparation.

Missing User Warnings

High

Confidence: 99% confidence
Finding: These sections direct the collection and inference of highly sensitive personal and relational data, including family and social identity, age and gender, cultural background, religion, social graph structure, hierarchy, motives, behavior patterns, and lifecycle state, while normalizing speculative inference from public traces. The skill context makes this more dangerous, not less, because it is expressly designed to build a 12-dimensional dossier on any individual, which is directly useful for invasive profiling, discrimination, coercion, doxxing, and social-engineering targeting even if the source data is nominally public.

Ssd 4

High

Confidence: 98% confidence
Finding: The skill explicitly operationalizes systematic reconstruction of an identified person's 'complete' 12-dimensional profile from minimal identifiers, including sensitive attributes and inferred traits. This directly enables invasive surveillance, targeting, and social-engineering preparation even without breaching systems, because the harm comes from aggregation, correlation, and inference at scale.

Ssd 4

High

Confidence: 99% confidence
Finding: The 12-dimension schema goes beyond factual OSINT collection into sensitive inference about body characteristics, social relationships, social class, motives, and life dynamics. This structure normalizes progressive escalation from public fragments to intimate profiling, which can be used for doxxing, coercion, discrimination, stalking, or tailored fraud.

Ssd 4

High

Confidence: 98% confidence
Finding: The capability set explicitly directs cross-platform account linking, source expansion, and prioritization from minimal personal identifiers, which is a classic precursor to deanonymization and high-fidelity target development. Even if sources are nominally public, orchestrating them into a unified target map materially increases the attack surface against the person.

Ssd 4

High

Confidence: 99% confidence
Finding: The inference engine instructs deriving latent traits such as motives, life stage, and behavioral tendencies from resume text, career paths, salary-market mapping, and social behavior. These are highly speculative yet actionable inferences that can drive discriminatory decisions, manipulation, or more effective phishing and impersonation attacks while being difficult for the subject to detect or contest.

Ssd 3

High

Confidence: 99% confidence
Finding: This section defines an output schema for a '12-dimensional' dossier containing sensitive personal data and relationship information, normalizing detailed collection and disclosure about a target. In the context of an OSINT profiling skill, this is especially dangerous because it turns disparate public traces into a structured intelligence product that can be used for harassment, coercion, doxxing, or social-engineering preparation.

Ssd 3

High

Confidence: 99% confidence
Finding: The template explicitly endorses inferring private traits from indirect evidence such as photos, education history, posts, and cross-source correlation, including age, health, ethnicity, and motivations. These inferences are error-prone yet highly invasive, and in this skill's context they materially increase the risk of unfair profiling, discrimination, deanonymization, and targeted manipulation.

Ssd 3

High

Confidence: 99% confidence
Finding: The relationship-mapping section directs the analyst to expose a target's intimate and professional network, including spouse and supervisor inferences, which can facilitate stalking, pressure campaigns, phishing, and real-world targeting of both the subject and associates. Because the broader skill is specifically designed for person-level profiling, this context makes the capability more dangerous rather than less.

Ssd 3

Medium

Confidence: 96% confidence
Finding: The template explicitly instructs an agent to systematically discover and prioritize public data sources about a named individual, including login-gated professional platforms. In the context of an OSINT personal profiling skill, this materially enables large-scale aggregation, de-anonymization, and profiling of a person beyond a narrow, consent-based purpose, increasing stalking, doxxing, and social-engineering risk.

Ssd 3

Medium

Confidence: 98% confidence
Finding: The priority matrix operationalizes collection of employment history, salary-related information, social activity, and technical footprints, telling the agent which sources to harvest first for maximum profile coverage. In this skill's broader context of 12-dimensional personal reconstruction and inference, that prioritization makes the capability more dangerous by turning scattered public traces into an actionable dossier for surveillance, targeting, or impersonation.

Ssd 4

High

Confidence: 99% confidence
Finding: The structured, multi-stage progression from minimal identifiers to a comprehensive 12-dimensional dossier meaningfully increases operational abuse potential by making invasive profiling repeatable and scalable. The detailed topology, iterative gap-filling, and confidence scoring normalize escalation and help an operator systematically deepen surveillance on a target individual.

Ssd 3

High

Confidence: 99% confidence
Finding: The instructions plainly direct the system to gather and output extensive sensitive data about a target person, creating a turnkey profiling capability. In context, this is dangerous because the skill is not merely descriptive; it prescribes actionable collection steps and output formatting for personal intelligence compilation.

Ssd 3

High

Confidence: 99% confidence
Finding: This section directs collection of intimate and protected attributes, including family identity, physical appearance, and cultural background, for inclusion in a final profile. Such aggregation can facilitate discrimination, targeting, and real-world harm, especially when combined with identity confirmation and cross-source correlation elsewhere in the workflow.

Ssd 3

Medium

Confidence: 97% confidence
Finding: Inferring motivation, decision patterns, and salary positioning from public traces is a form of psycho-economic profiling that can be inaccurate yet highly harmful when used for targeting or decision-making. In this skill's broader context, these inferences are combined with identity and social mapping, making manipulation and discriminatory use more feasible.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.