Back to skill

Security audit

Pet Grooming Effectiveness & Hairball Risk Analysis | 宠物梳毛器梳理效果与毛球风险

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because scanners report credential and identity handling, persistent user data and history access, and behaviors beyond its stated grooming analysis purpose.

Consider the suspicious verdict before installing. In particular, review whether you are comfortable with the reported credential and identity handling, persistent profile and report history access, and remote processing of local media. Scanning was partial because a referenced local path was missing, so the evidence is incomplete.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (26)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The supplied chunk is a user-account database component, not an image/video analysis implementation. Its User model includes email, birthday, sex, age, token, and open_token fields; its DAO supports reading, writing, and deleting these records. Database initialization also creates tables and adds missing user columns. Persistence can be a supporting implementation detail, but this chunk establishes account and credential management without any visible connection to grooming analysis or analysis-result storage. No media processing, server-side recognition request, or hairball-risk calculation appears here. These capabilities are implemented for invocation; the chunk does not itself instantiate the DAO or demonstrate token collection or transmission.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The analysis commands align with the declared purpose, and report formatting and output-file options are supporting details. However, historical-report retrieval and its explicit keyword triggers are not declared and operate independently of the described media-analysis trigger. The underlying Python implementation is not supplied, so no additional behavior or resource access can be established.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

The primary analysis workflow is consistent with the description, although the imported implementation is unavailable, so the specific recognition metrics and server interactions cannot be verified. The mismatch is the additional standalone history-query capability: --list bypasses media analysis and requests records associated with the current OpenID. This is a separate user-facing data-access operation rather than merely an implementation detail of analyzing supplied media. No unrelated data exfiltration is evident in the supplied chunk.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
87% confidence
Finding

The primary analysis workflow matches the declared purpose. However, --list provides a separate history-retrieval operation that requires no image or video and accesses records associated with a resolved user identity. This is an undeclared data-access capability rather than merely an implementation detail of analyzing supplied media. The imported analysis implementation is not shown, so its API behavior and reported measurements cannot be independently verified from this chunk.

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-grooming-effectiveness-analysis"
description: "Triggers when a user provides a pet grooming area video or image URL/file for analysis; supports local uploads or network URLs to call server-side APIs for coat condition and shed hair recognition, detecting matting area ratio and shed hair volume to output hairball risk level, helping prevent hairball syndrome. Application scenarios: smart grooming tools, long-haired pet care, pet health management. | 当用户提供梳毛器区域的视频/图像URL或文件时,触发本技能进行毛发表面状态分析;支持通过上传本地视频/图片或网络URL,调用服务端A

The HTTP helper can forward service credentials and user identity to arbitrary external destinations.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The helper accepts arbitrary absolute HTTP or HTTPS URLs and automatically attaches service tokens, the API key, and user identity without checking the destination. If an attacker-controlled media URL reaches this helper, those credentials can be sent directly to the attacker's host; accepting HTTP also permits cleartext exposure.

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.

Persistent default-identity reuse and report association lack a user-facing privacy notice.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The instructions silently reuse a persistent local default identity to associate cloud reports without requiring a user-facing explanation. In a shared installation, an inadequately scoped default identity could mix different users' reports or expose earlier records through history queries; backend isolation is not established here.

Content

No source excerpt is available for this finding.

The --export-env-only option promises not to run analysis, but the code ignores it and proceeds with analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The parser accepts --export-env-only and promises not to analyze, but execution never checks that flag. When supplied with media input, it proceeds into the analysis workflow despite the caller's explicit no-analysis selection, potentially causing unintended external processing.

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

Local input files are submitted to the analysis API without disclosure that their contents leave the system.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The code reads the entire local input file, packages its contents as a file payload, and passes it to the analysis API. There is no confirmation, user-facing notice, or comment explaining the remote transmission; the existing comment only describes reading the file.

Content

No source excerpt is available for this finding.

The --export-env-only option promises not to analyze media, but does not prevent analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The --export-env-only flag is parsed but never consulted, so providing it with a media input still reaches analyze_video. This violates an explicit no-analysis promise and can initiate unwanted media processing and API activity.

Content

No source excerpt is available for this finding.

The service exposes username-based user lookup, which is unrelated to grooming-media analysis.

Medium
Category
Excessive Agency
Confidence
88% confidence
Finding

Lines L098–L101 provide a user-lookup capability by forwarding a supplied username to RequestUtil.get_user_by_username. The manifest describes uploading grooming media and obtaining coat, shedding, and hairball-risk assessments; looking up user accounts is not a declared or obvious requirement of that workflow.

Content

No source excerpt is available for this finding.

The skill includes persistent user-profile and credential management unrelated to grooming-media analysis.

Medium
Category
Excessive Agency
Confidence
88% confidence
Finding

L448-L473 define a persistent user model containing real names, email addresses, birthdays, sex, age, and authentication tokens, and expose it through UserDao with inherited create, update, and delete operations. DAO initialization also creates tables and migrates sys_user fields (L184-L215). Although server-side API calls may require authentication, general-purpose personal-profile storage and mutation exceed the stated purpose of analyzing grooming images or videos.

Content

No source excerpt is available for this finding.

FileUtil.open silently truncates existing files without disclosing its destructive behavior.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

Opening a path in 'w' mode immediately erases any existing contents. This generic open helper has no docstring, explanatory comment, confirmation, or user-facing notification describing that behavior; its exception handler does not disclose successful truncation.

Content

No source excerpt is available for this finding.

The HTTP helper silently submits identity data for remote login or registration.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

An ordinary request can automatically trigger silent login or registration, transmitting the selected identity as both openId and mobile without a separate account-creation decision. This exceeds the advertised media-analysis behavior, and the caller can select API_SECRET_KEY as that identity, potentially disclosing a credential in identity fields and backend logs.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
93% confidence
Finding

External transmission is expected for server-side grooming analysis, but this particular POST sends identity data for silent login or registration rather than media analysis. It is the transmission sink for the automatic-registration issue above, including the possibility that an API key is reused as identity data; it does not establish malicious exfiltration to an unrelated domain.

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"):

Generic history-report phrases can trigger grooming-report queries outside the intended domain.

Low
Category
Trigger Abuse
Confidence
93% confidence
Finding

Examples include generic history-report language, while the instructions require automatic cloud queries when history keywords match. In an unrelated conversation, this can cause unintended retrieval and display of grooming records, although actual exposure depends on the agent's routing and account context.

Content

No source excerpt is available for this finding.

Configuration loading silently creates directories and files, including during module import.

Low
Category
Output Handling
Confidence
93% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested path does not exist, without a log, prompt, comment, or docstring disclosing this side effect. BaseEnum.init_subclass invokes this loader when the configuration classes are defined, so importing the module can modify the filesystem without an explicit save request.

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 353)May include surrounding context.

python
column_names = self.__model__.__table__.columns.keys()

            for key in column_names:
                value = getattr(model, key)
                setattr(instance, key, value)

            session.commit()

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)

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 441)May include surrounding context.

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

            return query.scalar()
        finally:

AgentSkill.ai_chat claims to invoke an agent and return analysis, but its implementation never invokes the agent.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L26–L34 states that this method invokes openclaw agent through subprocess and returns an analysis result or session ID. However, command execution is commented out, and the empty dictionary assigned at L68 causes result.stderr to raise an AttributeError, which is caught and suppressed; the method therefore returns None rather than performing the documented delegation.

Content

No source excerpt is available for this finding.