Back to skill

Security audit

Psychological Stress Assessment Skill | 心理压力评估技能

Security checks across malware telemetry and agentic risk

Overview

This skill mostly matches its stated stress-assessment purpose, but it handles sensitive biometric and mental-health data with automatic cloud history access and hidden account/token persistence that deserves review before installation.

Install only if you are comfortable sending face images/videos and derived psychological-stress assessments to the configured lifeemergence.com services. Before use, verify the provider, retention/deletion policy, consent requirements, and whether automatic history lookup and local token storage are acceptable in your environment.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (31)

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Content
if filters:
                for key, value in filters.items():
                    query = query.filter(getattr(self.__model__, key) == value)

            if offset:
                query = query.offset(offset)
Confidence
78% confidence
Finding
query = query.filter(getattr(self.__model__, key) == value)

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Content
if filters:
                for key, value in filters.items():
                    query = query.filter(getattr(self.__model__, key) == value)

            return query.scalar()
        finally:
Confidence
77% confidence
Finding
query = query.filter(getattr(self.__model__, key) == value)

Lp3

Medium
Category
MCP Least Privilege
Confidence
96% confidence
Finding
The skill instructs use of shell commands, local file handling, network access, and implicit environment/identity handling, yet declares no permissions or trust boundaries. This mismatch can cause operators or agent frameworks to execute capabilities without informed consent or policy gating, increasing the risk of unintended data access and exfiltration.

Description-Behavior Mismatch

Medium
Confidence
92% confidence
Finding
The documented behavior expands from psychological assessment to cloud-based historical report retrieval and report-link output, which materially broadens the skill's access to user data. That creates additional privacy and security exposure because historical report listing can reveal prior sensitive assessments and associated links beyond the user's immediate input.

Context-Inappropriate Capability

Medium
Confidence
95% confidence
Finding
Automatically creating and reusing a local default user identity introduces hidden state and implicit account linkage for highly sensitive mental-health and biometric data. If exploited or misconfigured, reports may be associated with the wrong person, enabling cross-user data leakage or unauthorized access to historical assessments.

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The documented API endpoints and scenario code describe a pet health analysis service, which materially conflicts with the skill's declared purpose of psychological stress assessment. This mismatch can cause the agent or integrator to call the wrong backend, mishandle sensitive health-related data, or expose unrelated reports/exports, especially because the documented export endpoint suggests access to full reports by ID.

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The skill’s stated purpose is media-based psychological stress assessment, but it also exposes a record-listing mode that enumerates prior assessments for a user. Because these are highly sensitive mental-health-related records and the capability is not clearly disclosed in the primary skill description, it expands scope into account data access and increases the risk of privacy violations or unauthorized history retrieval.

Context-Inappropriate Capability

Medium
Confidence
90% confidence
Finding
The code resolves and uses an internal user identity for account-scoped operations even though simple media analysis does not inherently require hidden identity linkage. In a mental health context, binding analyses to internal identifiers creates sensitive profiling and privacy risks, especially if users are not clearly informed or if identity resolution occurs implicitly.

Description-Behavior Mismatch

Low
Confidence
88% confidence
Finding
The skill accepts arbitrary HTTP/HTTPS URLs and forwards them for analysis, even though the declared purpose is psychological stress assessment from video content rather than general URL retrieval. This expands the capability surface and can enable unintended remote resource access, privacy issues, or backend abuse if the downstream service fetches attacker-controlled URLs.

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The skill exposes report-listing and export-link generation features that are not described in the manifest, creating hidden data access functionality beyond simple analysis. In a mental-health context, historical report access and export URLs may reveal sensitive psychological assessment data and increase the risk of unauthorized disclosure.

Context-Inappropriate Capability

Medium
Confidence
95% confidence
Finding
Historical report enumeration is not necessary for the stated task of analyzing a supplied video and materially increases exposure of prior assessment records. Because the domain is psychological stress, anxiety, and depression monitoring, enumerating past reports can leak highly sensitive health-related information if authorization boundaries are weak or absent.

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
This file exposes broad generic HTTP and CRUD wrapper methods (add, edit, delete, http_get/post/put/delete) that can interact with arbitrary URLs, which materially exceeds a skill described as performing psychological stress assessment. In an agent setting, these capabilities can be repurposed to access or modify unrelated remote resources, increasing the attack surface and enabling misuse beyond the declared purpose.

Context-Inappropriate Capability

High
Confidence
97% confidence
Finding
The add, edit, and delete methods provide arbitrary remote resource modification via POST requests without visible restrictions tied to the mental-health analysis use case. If exposed to agent-controlled inputs, this can enable unauthorized state changes on backend services or connected systems, which is especially risky because the capability is unrelated to the manifest-stated purpose.

Context-Inappropriate Capability

Medium
Confidence
88% confidence
Finding
The get_user_by_username helper introduces account lookup functionality that is not justified by a stress-assessment analysis skill. This can facilitate user enumeration or retrieval of identity-linked information, creating unnecessary privacy and reconnaissance risk in a mental-health context where data sensitivity is high.

Context-Inappropriate Capability

Medium
Confidence
88% confidence
Finding
The config layer reads environment variables containing user identifiers such as OPENCLAW_SENDER_OPEN_ID, OPENCLAW_SENDER_USERNAME, and FEISHU_OPEN_ID, then stores them globally without any purpose limitation or disclosure in this file. In a psychological stress assessment skill, silently ingesting identity data expands data collection beyond what is necessary for the stated function and increases privacy and misuse risk if other components log, transmit, or repurpose these values.

Context-Inappropriate Capability

Medium
Confidence
87% confidence
Finding
The file defines storage and retrieval of user identity/profile records including username, realname, email, birthday, age, sex, and tokens, which are unrelated to the stated psychological stress assessment purpose. In a mental-health monitoring context this widens collection of sensitive personal data and creates unnecessary privacy and compliance risk, especially when combined with shared local persistence.

Description-Behavior Mismatch

Medium
Confidence
85% confidence
Finding
This module implements a shared local SQLite DAO, path redirection, and schema migration for user/account data rather than stress-analysis logic described in the manifest. The mismatch is dangerous because it indicates hidden persistence and cross-skill data handling in a mental-health skill, increasing the chance of undisclosed tracking, data mixing between contexts, and overcollection of sensitive information.

Context-Inappropriate Capability

High
Confidence
97% confidence
Finding
This utility provisions, resolves, and persists user identities and tokens locally, including generating default identities and reading identity material from workspace files, which exceeds the stated purpose of a psychological stress assessment skill. In a mental-health context, silent identity management increases the privacy risk because sensitive assessment activity can become linked to durable user identifiers without clear consent or necessity.

Context-Inappropriate Capability

High
Confidence
98% confidence
Finding
The HTTP wrapper silently auto-registers or logs in users against remote health endpoints when tokens are missing, then stores returned tokens for reuse. That is an unjustified account lifecycle capability for a stress-analysis skill and is especially sensitive here because the skill domain involves mental-health data, making undisclosed identity linkage and remote transmission more dangerous.

Description-Behavior Mismatch

Medium
Confidence
89% confidence
Finding
The file contains broad workspace discovery and directory creation logic unrelated to a narrowly described stress-assessment function. While not inherently malicious, this expands the skill's operating scope across agent workspaces and local storage, increasing the blast radius for data access or unintended persistence beyond what users would reasonably expect from the manifest.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The auto-trigger phrases for historical report retrieval are broad enough to match ordinary conversation, which can cause unintended cloud queries for sensitive mental-health history. In this context, accidental invocation is especially risky because the returned data concerns biometric and psychological assessment records.

Missing User Warnings

High
Confidence
98% confidence
Finding
The skill processes face images/videos and queries cloud-hosted historical reports, but it does not present a clear privacy and data-handling warning for biometric and mental-health data. This omission undermines informed consent and increases the chance that users or integrators will expose extremely sensitive personal information without understanding storage, transfer, retention, or sharing implications.

Missing User Warnings

High
Confidence
96% confidence
Finding
The skill sends media for psychological assessment to an external API without an explicit user-facing warning that highly sensitive biometric and mental-health-related data will leave the local environment. In this context, lack of clear disclosure and consent is especially dangerous because users may unknowingly transmit intimate health inferences, creating legal, privacy, and trust risks.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The delete method triggers a network request directly with no visible confirmation, safety interlock, or indication that the target is constrained to safe resources. In an agent environment, silent destructive actions can be invoked programmatically and may delete or alter remote data without meaningful user awareness or validation.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The code accesses user-identifying environment variables and populates shared runtime state without any visible notice, consent flow, or minimization controls. For a mental-health-related skill, undisclosed handling of identifiers is more sensitive because identity linked to stress/anxiety/depression analysis can create significant privacy and profiling concerns.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

Detected: suspicious.install_untrusted_source

Install source points to URL shortener or raw IP.

Warn
Code
suspicious.install_untrusted_source
Location
skills/smyx_common/scripts/config-dev.yaml:2