Back to skill

Security audit

Pet UV Safety Monitor Analysis | 宠物紫外线消毒安全监测

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because its reported account, credential, and video-analysis behavior does not clearly match the advertised live pet and UV-C safety monitoring, and it has notable privacy and side-effect concerns.

Consider avoiding installation until the account and credential handling, external video processing, and environment-only behavior are clearly explained and addressed. The scanners did not establish malicious intent, and their assessment is partial because a referenced local path was missing.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
Findings (28)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied code primarily manages a sys_user database containing usernames, real names, email addresses, demographics, and token fields. Generic database storage could support the declared event logging, but this chunk specifically implements user-account persistence and contains no UV-safety event model or monitoring logic. Its demonstrated primary purpose therefore materially differs from the description. This assessment is limited to the supplied chunk; it does not establish what other files implement.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied code is generic, nonfunctional agent-call scaffolding rather than the declared pet UV-safety monitor. It does not execute the commented-out subprocess command or demonstrate any of the core monitoring and protection workflow. Imported base classes may provide additional functionality, but their implementations are not supplied and cannot establish that the declared behavior exists.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

Generic file, JSON, polling, and HTTP helpers could legitimately support the declared monitor and are not mismatches by themselves. However, the supplied chunk implements a substantial identity, remote-account, authentication, and billing workflow not disclosed in the description, with no visible connection to UV safety monitoring. The remote login payload explicitly requests silent registration and transmits the selected identity. This assessment is limited to the supplied chunk; it does not establish whether monitoring functionality exists elsewhere or whether these methods are invoked.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The manifest promises real-time monitoring and automatic protective actions, while its operational workflow describes submitted-file analysis and report retrieval without establishing a continuous monitoring or lamp-control path. In a UV-C safety application, users relying on those guarantees could leave pets exposed while believing protection is active.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The real-time safety claims exceed the submitted-video and reporting workflow established by this document, creating potentially harmful reliance on unverified protection. However, history retrieval is explicitly documented and is not an undisclosed capability.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

Generic AI-analysis endpoints could support the declared monitoring system and are not inherently mismatched. However, the health-order report export endpoint is an inconsistent resource not explained by the description. This chunk only configures URLs; it does not demonstrate actual network requests or report exports. The absence of monitoring logic in this configuration file alone would not establish a mismatch.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
87% confidence
Finding

The supplied code implements a generic API access layer rather than the described pet UV-safety monitor. HTTP helpers could legitimately support that monitor, so their presence alone is not suspicious; however, this chunk provides no connection between its operations and the declared safety functionality, and exposes unrelated user-lookup and download-link functionality. This assessment is limited to the supplied chunk: it does not establish that other files lack the monitoring implementation or that these methods are actually invoked.

Content

No source excerpt is available for this finding.

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: "smyx-uv-safety-monitor-analysis"
description: "AI-powered UV disinfection safety monitor for pets. Real-time camera analysis detects whether a pet enters an active UV-C disinfection zone and whether the UV lamp is on (via blue-purple glow recognition or smart-home API linkage). When both conditions are met, it auto-triggers a high-risk alert, recommends shutting off the UV lamp, and logs the event to prevent corneal burns or skin damage. Scenarios: smart homes, pet households, pet boarding facilities. | 通过智能家居摄像头实时识别宠物是否进入正在进行紫外线消毒的区域,自动关闭UV灯并推送�

The request helper attaches shared backend credentials and identity to arbitrary destination URLs.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The helper accepts arbitrary absolute HTTP or HTTPS URLs and attaches shared access tokens, API keys, authorization tokens, and identity metadata without checking the destination. A caller-influenced URL can disclose credentials to an attacker, while an HTTP destination also exposes them in transit; smart-home integrations make destination separation particularly important.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The local-video analysis workflow lacks a privacy warning that indoor camera footage is sent to an external API for processing.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

The workflow directs analysis of local indoor footage without clearly warning that the footage will be uploaded for external processing. General API and cloud-report references do not adequately communicate that private household video itself leaves the device, creating a meaningful privacy risk.

Content

No source excerpt is available for this finding.

The --export-env-only help promises no analysis (L53–L54), but the flag is ignored and analysis still executes when an input is supplied (L68–L80).

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The parser accepts --export-env-only and explicitly promises not to analyze, but the code never checks the flag before identity initialization and analysis. Supplying an input can therefore submit private footage and potentially consume service resources despite the caller selecting a non-analysis mode.

Content

No source excerpt is available for this finding.

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · skills/smyx_analysis/scripts/skill.py (reported line 28)May include surrounding context.

python
result_json = JsonUtil.parse(result_json_pure_text, result_json_pure_text)

        result_json_common_ai_response = result_json.get("commonAiResponse") if isinstance(result_json,
                                                                                           dict) else result_json
        if result_json_common_ai_response:
            result_json = result_json_common_ai_response

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · skills/smyx_analysis/scripts/skill.py (reported line 33)May include surrounding context.

python
result_json = result_json_common_ai_response

        result_json_health_ai_response = result_json.get("healthAiResponse") if isinstance(result_json,
                                                                                           dict) else result_json
        if result_json_health_ai_response:
            result_json = result_json_health_ai_response

The documented environment-only mode still executes video analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The environment-only flag is parsed but never consulted, so an invocation with video input proceeds to analysis despite the explicit no-analysis promise. This can cause unintended external processing of private footage and service usage.

Content

No source excerpt is available for this finding.

Database initialization unconditionally modifies an unrelated system-user schema.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

Every Dao initialization calls _alter_tables() at L186, which inspects sys_user and adds source_id and realname columns if missing. Modifying a system-user identity schema is not necessary for the declared UV monitoring, lamp control, alerting, or event logging functions, and this migration runs regardless of the DAO's selected model.

Content

No source excerpt is available for this finding.

save() is documented as creating a new record but can overwrite an existing record.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

Any insertion exception triggers attempts to overwrite an existing record by ID and then by username or realname, copying fields including credentials and identifiers. A conflicting or malformed supplied record can therefore turn an expected create operation into unintended modification of another identity, although exploitation depends on control over supplied models.

Content

No source excerpt is available for this finding.

The ai_chat docstring claims to invoke the OpenClaw agent, but the implementation never executes it and cannot return an analysis result.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L25–L34 explicitly promises a subprocess-based OpenClaw invocation that returns an analysis result or session ID. However, the invocation is commented out, and L68 initializes a dictionary whose nonexistent stderr attribute is accessed at L70; the resulting exception is caught and suppressed, so the method returns None without performing analysis. If used for the advertised safety monitoring, this method would provide no AI analysis despite its documented contract.

Content

No source excerpt is available for this finding.

FileUtil.open silently truncates existing files without disclosing its destructive write mode.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

At L101, opening a path with mode 'w' immediately erases any existing contents. The generically named helper provides no docstring, comment, confirmation, or user-facing log explaining this destructive behavior; exception handling does not disclose successful truncation.

Content

No source excerpt is available for this finding.

HTTP requests can silently register an account with a separate health-service backend.

Medium
Category
Excessive Agency
Confidence
95% confidence
Finding

When credentials are unavailable, an ordinary request can send the current identity to a health-service login endpoint with silent registration enabled. This introduces an undisclosed account-creation and identity-sharing side effect that is not inherent to camera-based UV safety monitoring.

Content

No source excerpt is available for this finding.

Ordinary HTTP requests can silently trigger remote account registration and local credential persistence.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

A routine HTTP operation can implicitly register a remote identity and persist returned token and openToken values in the local user database. These hidden side effects expand identity exposure and leave reusable credentials at rest without an explicit enrollment or credential-storage decision.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
89% confidence
Finding

This transmission sends the selected identity as both openId and mobile to a separate health-service endpoint during automatic login or registration. The issue is the implicit identity disclosure, not network access alone; the supplied code does not establish a malicious destination or show camera frames being transmitted here.

Content

Scanner excerpt · skills/smyx_common/scripts/util.py (reported line 561)May include surrounding context.

python
"source": ConstantEnum.DEFAULT__SKILL_HUB_NAME
            }
            try:
                _response = requests.post(_url, json=_data)
                if _response.status_code == 200:
                    _response_json = _response.json()
                    if _response_json and _response_json.get("success"):

Loading configuration silently creates directories and files, including during module import.

Low
Category
Output Handling
Confidence
92% confidence
Finding

When a configuration file is absent, YamlUtil.load creates its parent directories and writes the file without a warning, log, docstring, or explanatory comment. BaseEnum.init_subclass calls this loader during class definition (L052–L058), so merely importing the module can modify the filesystem rather than only read configuration.

Content

No source excerpt is available for this finding.

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · skills/smyx_common/scripts/dao.py (reported line 243)May include surrounding context.

python
username = getattr(model, "username", None)
            if username:
                column_names = self.__model__.__table__.columns.keys()
                update_data = {key: getattr(model, key) for key in column_names if key != "username"}
                return self.update_by_username(username, **update_data)

            return None

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · skills/smyx_common/scripts/dao.py (reported line 326)May include surrounding context.

python
if filters:
                for key, value in filters.items():
                    query = query.filter(getattr(self.__model__, key) == value)

            if offset:
                query = query.offset(offset)