Back to skill

Security audit

Elderly Facial Asymmetry / Mouth-Corner Deviation Detection | 老年人面部不对称/口角歪斜识别

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because scanners found undeclared account, credential, and network behavior, plus a flag that may allow analysis to run when users expect it not to.

The verdict is suspicious, so consider the reported account provisioning, credential storage, and possible unexpected media processing before installing. The scan was marked partial because a local reference could not be resolved; despite all listed components being scanned, the evidence is not complete, and this review does not establish that the skill is safe.

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 (22)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

Ordinary file, JSON, date, and polling helpers could support the declared analysis and are not independently concerning. However, persistent account provisioning, credential handling, and payment integration are substantial capabilities absent from the description and declared permissions. The request wrapper also attaches credentials without restricting absolute URLs to the configured service, and debug mode enables HTTP-level logging that may expose sensitive traffic. No camera capture, landmark detection, facial-feature measurement, or asymmetry scoring appears in this chunk. Because this is a shared utility module, their absence does not establish that the full skill lacks those features; the mismatch is based primarily on the concrete undeclared account and network behavior.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied chunk does not implement the declared facial-screening workflow; its substantive visible logic is an inactive and broken generic chat helper. No external agent command actually executes. Imported base classes and API services are not supplied, so their behavior cannot be assessed; the mismatch finding is limited to the visible code.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied code’s concrete functionality is generic account-data persistence rather than facial analysis. Although database utilities can support a larger application, this chunk manages user identities and credential fields without any demonstrated connection to facial-screening results. That is a materially different capability from the declared purpose. This assessment is limited to the supplied chunk; it does not establish what other skill files implement or imply that stored data is transmitted externally.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
89% confidence
Finding

The visible code implements generic backend API access rather than the declared facial-analysis capability. HTTP helpers alone could be supporting implementation details, but the user-lookup and general management interfaces have no demonstrated relationship to the stated task. This assessment is limited to the supplied chunk: it does not establish that these methods are invoked, that sensitive data is transmitted, or that facial-analysis functionality is absent elsewhere.

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-elderly-facial-asymmetry-analysis"
description: "Using a fixed home camera to capture frontal facial images or short videos of the elderly, the system uses AI facial-landmark detection to analyze features such as the height difference between left/right mouth corners, the symmetry of nasolabial folds (smile lines), and the asymmetry of eyebrow lifts, and computes a facial asymmetry index (0-100%). | 通过家庭固定摄像头拍摄老年人正面面部图像或视频,利用AI面部关键点检测技术分析左右嘴角的高度差、鼻唇沟(法令纹)的对称性、眉毛抬高的差异等特征,计算面部不对�

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 --export-env-only option promises not to execute analysis, but the program ignores it and follows the normal analysis path.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The program parses --export-env-only but never checks it, so providing media with this ostensibly non-analysis option still reaches the normal analysis call. This can unexpectedly transmit sensitive facial imagery for processing, and identity initialization also occurs before any such safeguard.

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 --export-env-only option promises not to analyze media, but supplying it still allows analysis to run.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The export-only flag is accepted but never used to stop execution, allowing supplied media to reach the normal analysis path. In this health-monitoring context, users relying on the explicit no-analysis promise could unintentionally disclose biometric or health-related imagery.

Content

No source excerpt is available for this finding.

The shared DAO provides persistent user-account and authentication-token management beyond the facial-asymmetry analysis scope.

Medium
Category
Excessive Agency
Confidence
86% confidence
Finding

The User model stores account identifiers, email addresses, and authentication material in token and open_token fields (L448-L465), and UserDao exposes the inherited database CRUD operations for those records (L470-L473). Managing persistent account credentials is not an obvious requirement of detecting facial landmarks or calculating asymmetry, and the manifest does not declare account or credential management.

Content

No source excerpt is available for this finding.

FileUtil.open silently truncates existing files without disclosing its write-only behavior.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

Opening a path with mode 'w' immediately erases any existing contents, even before the caller writes anything. This generically named helper has no docstring, comment, confirmation, or user-facing message explaining that destructive behavior; exception tracing does not disclose successful truncation.

Content

No source excerpt is available for this finding.

The request helper silently registers remote accounts and persists account credentials, beyond the declared facial-asymmetry analysis workflow.

Medium
Category
Excessive Agency
Confidence
90% confidence
Finding

When tokens are missing, a generic request can silently register or log in a remote account using the current identity and save returned account information and tokens locally. This undisclosed provisioning creates privacy and credential-lifecycle risks, particularly for an elderly-health screening skill; the supplied code does not establish how stored credentials are protected.

Content

No source excerpt is available for this finding.

An ordinary HTTP request can silently trigger remote account registration.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The helper sends silent=1 and register=1 to the remote login endpoint as an implicit consequence of making an ordinary request. This bypasses an explicit account-creation decision and can disclose the configured identity without an informed provisioning step; it is the same underlying behavior covered by the broader account-persistence finding.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
87% confidence
Finding

This POST transmits the selected username as both openId and mobile during implicit remote provisioning, without a consent gate in the supplied code. The selected value can also be API_SECRET_KEY, potentially placing a credential in identity fields subject to server-side logging or storage; there is no evidence that this particular request transmits facial images.

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"):

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

Low
Category
Output Handling
Confidence
94% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested configuration is missing, without a log, confirmation, docstring, or explanatory comment. BaseEnum.init_subclass invokes this loader during class definitions (L044–L059), so merely importing the module can modify the filesystem without disclosing that side effect.

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:

The ai_chat docstring claims agent execution, but the implementation never invokes an agent and cannot return an analysis result.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L25–L34 states that this method invokes the openclaw agent through subprocess and returns an analysis result or session ID. However, execution is commented out at L61–L66, and the empty dictionary assigned at L68 causes result.stderr at L70 to raise AttributeError, which is caught and suppressed. Consequently, the method returns None rather than performing its documented AI analysis.

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/util.py (reported line 30)May include surrounding context.

python
elif isinstance(cur, dict):
                        cur = cur.get(part, default)
                    else:
                        cur = getattr(cur, part)
                return cur
            except Exception:
                return default