Back to skill

Security audit

Child Bedtime Soothing (Fear of Dark / Post-Nightmare) | 儿童睡前情绪安抚(怕黑/噩梦后)

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because scanners report a mismatch between its child bedtime monitoring claims and the demonstrated capabilities, along with privacy, credential, and unintended-action concerns.

Consider avoiding installation unless you can verify the remote service, understand what recordings and identity data it receives, and confirm that any export-only action cannot process media. The scanner says coverage is partial, so these results do not establish that the artifact is safe or that the claimed monitoring and soothing functions work as described.

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
skills/smyx_common/scripts/config-dev.yaml
Finding

Development API configuration exposes credentials and private media to network interception

Content
View full analysis

Impact

An on-path attacker can capture service tokens and private media uploads or media URLs, read returned reports, and alter API responses. Captured bearer tokens can be replayed while valid.

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
98% confidence
Finding

The supplied chunk implements account, authentication, workspace, and API infrastructure rather than child-distress detection or soothing-device control. It contains no camera or microphone ingestion, audiovisual behavior analysis, nightlight activation, or audio playback. Generic helpers alone would not establish a mismatch, but the concrete remote account-registration and persistent identity-management behavior is not represented in the declared purpose or connected to that workflow in the supplied code. This conclusion is limited to the provided chunk; it does not establish what other skill files implement or imply malicious intent.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied code's concrete purpose is user-database management rather than child monitoring or automated soothing. Database helpers could support a larger monitoring application, but this chunk specifically manages usernames, real names, email addresses, demographics, and tokens without any visible connection to the declared functionality. No network transmission or sensor access is shown. This assessment is limited to the supplied chunk.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied code does not substantiate the declared primary purpose. It contains neither the monitoring and detection logic nor the automatic soothing actions described. This assessment is limited to the supplied chunk: inherited classes and imported modules are not shown and could contain additional functionality. The commented-out agent command is not treated as an executed capability.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

History retrieval is disclosed in the full skill documentation and is not itself suspicious. However, the automatic monitoring and soothing claims exceed the demonstrated file/URL analysis workflow, creating a material risk of misplaced caregiver reliance.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The skill promises automatic nighttime monitoring, soothing, and escalation, but the supplied execution path demonstrates only submitted-media analysis and report retrieval. In a child-monitoring context, presenting these unverified capabilities as operational can cause caregivers to rely on alerts or interventions that never occur.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

The supplied chunk implements backend API utilities rather than camera or microphone analysis, distress detection, or soothing-device control. HTTP helpers could legitimately support the declared system, but the storage-download and user-lookup capabilities have no demonstrated relationship to that purpose. This assessment is limited to the supplied chunk: it does not establish that these methods are invoked, that data is exfiltrated, or that the complete skill lacks the declared functionality.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
83% confidence
Finding

Generic AI-analysis configuration could support the declared purpose, and missing detection logic in a configuration-only chunk is not by itself a mismatch. However, the explicit health-order report export resource is outside the described workflow. This chunk only configures that URL; it does not demonstrate an actual report request, data export, or network transmission.

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-child-bedtime-soothing-analysis"
description: "Through a fixed camera (with infrared night vision) and microphone in the child's bedroom, the system analyzes pre-sleep and night-time video and audio to detect pre-sleep crying (continuous crying, calling 'Mama'), fear-of-the-dark expressions (curling up, looking around), and nightmare awakenings (sudden sitting up, trembling, screaming). | 通过儿童卧室的固定摄像头(红外夜视)及麦克风,分析儿童睡前及夜间视频,检测睡前哭闹(持续性哭声、呼喊'妈妈')、怕黑表现(身体蜷缩、四处张望)、噩梦惊醒(突然坐起�

The analysis workflow lacks a clear privacy warning that children's bedroom recordings are processed by a remote API.

High
Category
Output Handling
Confidence
95% confidence
Finding

Although the documentation mentions APIs and cloud reports, it does not prominently explain that local bedroom recordings will be transmitted to a remote processor or require informed approval before transmission. Children's intimate audio and video make this a particularly sensitive privacy boundary, and stated retention restrictions are not backed by visible enforcement.

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 --export-env-only help promises no analysis, but the flag is ignored and analysis still runs.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The parser accepts --export-env-only and promises no analysis, but main never checks the flag. With an input supplied, execution proceeds through identity initialization and analysis, potentially uploading a child's bedroom recording despite an explicit request for a non-analysis operation.

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 disclosing the upload to the user.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The code reads the complete local file and passes its bytes to the analysis service without a visible disclosure or consent check. In the stated use case, this crosses a sensitive privacy boundary by transmitting children's bedroom audio and video rather than merely processing them locally.

Content

No source excerpt is available for this finding.

The documented export-only mode still executes video analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The export-only flag is parsed but never consulted, so an invocation with media input still initializes identity and executes analysis. Users expecting a non-processing operation can inadvertently transmit sensitive recordings, with elevated consequences when this utility handles children's bedroom footage.

Content

No source excerpt is available for this finding.

The service exposes username-based user lookup unrelated to bedtime analysis or soothing.

Medium
Category
Excessive Agency
Confidence
85% confidence
Finding

At L098–L101, get_user_by_username delegates lookup of a caller-supplied username to RequestUtil. User-directory lookup is not a direct requirement of detecting childhood distress or controlling soothing devices, and the manifest does not declare account-search functionality.

Content

No source excerpt is available for this finding.

The skill includes system-user and credential management unrelated to bedtime monitoring and soothing (L448–L473).

Medium
Category
Excessive Agency
Confidence
90% confidence
Finding

The User model stores account identifiers, personal information, token, and open_token, and UserDao exposes inherited read, update, and delete operations on these records. DAO initialization also creates registered tables and explicitly migrates sys_user (L180–L215), making this more than an unused schema declaration. Managing system accounts and credentials is not a direct requirement of the described audiovisual distress detection and soothing actions.

Content

No source excerpt is available for this finding.

ai_chat documents an agent invocation, but the implementation never invokes the agent.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L24–L35 states that ai_chat invokes the openclaw agent through subprocess and returns an analysis result or session ID. However, the subprocess invocation is commented out at L61–L66; the replacement empty dictionary at L68 causes result.stderr at L70 to raise an exception, which is caught without returning an analysis result. This materially contradicts the documented delegation of analysis rather than merely changing an implementation detail.

Content

No source excerpt is available for this finding.

The generic file-opening helper silently truncates existing files.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

FileUtil.open always opens the supplied path in write mode, immediately erasing existing contents even before anything is written. Its generic name does not disclose this destructive behavior, and there is no explanatory comment, docstring, confirmation, or action-specific log.

Content

No source excerpt is available for this finding.

The request helper silently registers or logs in an external service account, beyond the stated bedtime-monitoring and soothing purpose.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

An ordinary HTTP request can implicitly submit an identity to an external phoneLogin endpoint with registration enabled, without a separate provisioning decision. This undisclosed account creation crosses a privacy and authorization boundary, particularly for a skill associated with children's bedroom monitoring; the supplied metadata does not disclose this dependency.

Content

No source excerpt is available for this finding.

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

Medium
Category
Output Handling
Confidence
93% confidence
Finding

The generic request path can register a remote account and persist its returned authentication tokens in a local SQLite user record, making a routine network operation perform hidden identity and credential-management writes. No explicit provisioning approval or credential-storage protection is shown, increasing privacy and credential-exposure concerns on shared agent workspaces.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
95% confidence
Finding

The POST transmits identity information to the configured health-service endpoint, and its caller prioritizes API_SECRET_KEY as the username, allowing that secret to be copied into both openId and mobile registration fields. The destination is not constrained to HTTPS or a trusted origin, so this can expose credentials through an unintended API field or insecure configuration; this call does not itself demonstrate transmission of bedroom recordings.

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 writes a new configuration file and creates parent directories when the requested path is absent, without a log, confirmation, docstring, or explanatory comment disclosing this side effect. BaseEnum.init_subclass invokes this loader as classes are defined (L044–L058), so merely importing this module can modify the filesystem.

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)