Back to skill

Security audit

smyx_infant_cry_cause_classification_analysis | 婴幼儿哭声原因分类

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because scanners report account and identity handling, remote media processing, and behavior that does not clearly match its infant-cry analysis description.

Consider the suspicious verdict and privacy concerns before installing. In particular, check whether you are comfortable with the reported remote handling of media and account registration, and whether the claimed purpose matches the behavior you need. Skillspector coverage was partial because a referenced local path was missing, so the available evidence does not establish the behavior of that reference.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/smyx_infant_cry_cause_classification_analysis.py
Finding

Default identity resolution discards the current sender identity and can reuse another user's report credentials

Content
View full analysis

Impact

In a shared-sender deployment, another sender can obtain infant-analysis history and associated report links through the reused account. This does not establish a vulnerability in the remote API: the client itself selects the wrong account's valid credentials.

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
Findings (26)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied code's substantive purpose is general database and user-account persistence, not infant-cry analysis. Database storage could support a larger monitoring application, but this chunk specifically manages user identity fields and tokens without any demonstrated connection to cry audio or analysis results. Under the supplied criteria, its materially different purpose and resource use constitute a mismatch. This assessment is limited to the supplied chunk; it does not establish what other skill components implement or imply malicious behavior.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied chunk does not substantiate the declared infant-monitoring purpose. Its visible behavior consists of generic scaffolding and a nonfunctional chat helper. There is no active subprocess execution or demonstrated unauthorized resource access. Imported base and API implementations are unavailable, so this assessment is limited to the supplied code.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

Ordinary formatting, polling, and HTTP helpers could support the declared skill and are not themselves mismatches. However, this chunk also implements an account-provisioning and identity-persistence workflow that is not disclosed in the infant-cry description. It contains no microphone capture, acoustic analysis, cry classification, confidence estimation, or parent-notification implementation. Because only a shared utility chunk is supplied, those features could exist elsewhere; nevertheless, the account-management behavior shown here materially exceeds the declared purpose. These actions occur when the relevant methods are called, not simply when the module is imported.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied code explicitly presents itself as a video analysis tool and implements video submission and history-listing workflows. It contains no visible microphone capture, continuous cry detection, acoustic feature extraction, cry-cause classification, confidence scoring, or mobile push notifications. The delegated analysis implementation is not supplied, so it could potentially analyze infant cries within videos; however, that possibility does not establish the declared functionality, and the visible video-history capability is undeclared.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied code implements an API-backed media-analysis and report-management interface. Its inputs are existing local files or video URLs, and its outputs are generic commonAiResponse or healthAiResponse reports. Nothing in the chunk establishes infant-cry analysis, microphone integration, real-time monitoring, or parent-app notifications. The imported API implementation could perform additional analysis, but that behavior is not demonstrated here. The visible workflow therefore materially differs from the declared purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The visible implementation is a general-purpose backend API utility rather than an infant-cry analysis service. HTTP helpers could support the declared system and are not inherently suspicious, but this chunk exposes unrelated operations and does not show how they support the stated purpose. This assessment is limited to the supplied code; it does not establish that the broader skill lacks cry-analysis functionality or that any exposed operation is 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-infant-cry-cause-classification-analysis"
description: "Using the built-in microphone of a baby monitor or smart camera to capture infant cry audio, AI acoustic analysis extracts cry features such as frequency, pitch, rhythm, and duration, and classifies the possible causes behind the cry (hunger, sleepiness, pain/discomfort, boredom/need for comfort, fear, etc.), outputting the most likely cause and its confidence. | 通过婴儿监护器或智能摄像头的内置麦克风采集婴儿哭声音频,利用AI声学分析技术提取哭声的频率、音调、节奏、持续时间等特征,分类识别婴儿哭声背后的可能原�

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 analysis workflow lacks a specific disclosure about remote processing of sensitive infant recordings.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow automatically submits recordings for API processing without clearly stating that local infant recordings leave the device or identifying the recipient and retention practices. Generic guardian-consent language and references to APIs do not adequately establish informed consent for transferring sensitive recordings of minors.

Content

No source excerpt is available for this finding.

The --export-env-only option promises not to analyze audio, 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 no analysis, but main never checks the flag before identity initialization or analysis. When an input accompanies the flag, a user expecting a non-processing operation can instead trigger submission of sensitive infant media.

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 video contents are passed to the analysis API without disclosure that the file will be uploaded.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

The code reads the entire local recording and passes its contents to the analysis API as a file attachment without a visible remote-processing notice or authorization check. For infant recordings, an unexpected upload can expose sensitive household audio and video involving minors.

Content

No source excerpt is available for this finding.

The --export-env-only option promises not to perform analysis, but does not prevent analysis from running.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The --export-env-only flag is accepted but never used to stop identity initialization, history requests, or analysis. Supplying media with this purported non-analysis mode can therefore initiate unintended remote processing, with heightened privacy consequences for infant recordings.

Content

No source excerpt is available for this finding.

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

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

Any insertion exception triggers an update by ID, followed by a fallback matching either username or realname and copying other columns, including identity and token fields. If an untrusted or mismatched model reaches this method, an operation presented as creation can overwrite another local user's data; the impact on sensitive family accounts depends on caller validation and record control.

Content

No source excerpt is available for this finding.

ai_chat claims to invoke an analysis agent, but execution is disabled and no analysis result is returned.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L25–L34 states that this method invokes openclaw agent through subprocess and returns an analysis result or session ID. In reality, the subprocess call is commented out, and accessing stderr on the empty dictionary at L68 raises an exception that is caught without returning a result. This contradicts the documented analysis behavior: the prompt is never submitted to an agent.

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

L101 opens the supplied path with mode 'w', immediately erasing existing contents even before anything is written. The generically named helper has no docstring, comment, confirmation, or success-path logging explaining this behavior; exception tracing does not disclose successful truncation.

Content

No source excerpt is available for this finding.

The HTTP helper silently registers a backend account, although the manifest only describes infant-cry analysis and notifications.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

When credentials are missing, the request helper can contact phoneLogin with register=1 and silent=1, transmitting an identity and requesting account creation without an explicit registration gate. Backend authentication may support notifications, but silently enrolling a user is an undisclosed side effect, particularly concerning in a service involving infants and family information.

Content

No source excerpt is available for this finding.

HTTP requests can silently trigger remote user registration without disclosure.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

An ordinary HTTP operation can trigger a separate remote registration request when tokens are absent, without an explicit caller option or user confirmation. This is the same underlying provisioning issue identified by SDI-1, and its hidden state-changing behavior creates privacy and account-management risks beyond the advertised cry-analysis operation.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

The registration POST sends the selected identity as both openId and mobile without an explicit consent gate; that identity can originate from API_SECRET_KEY, causing a configured secret to be copied into identity fields. The source does not establish a malicious destination or transmission of infant audio here, but the unnecessary disclosure and credential repurposing are concrete concerns.

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 writes missing configuration files.

Low
Category
Output Handling
Confidence
94% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested path is missing, without a prompt, log, docstring, or explanatory comment disclosing this side effect. Because BaseEnum.init_subclass invokes this loader at L053 and L058, merely defining configuration subclasses can cause filesystem changes rather than only reading 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 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: