Back to skill

Security audit

Child Social Interaction Frequency & Duration Analysis | 儿童社交互动频次与时长分析

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because scanners report undeclared account and credential handling, possible transfer of children’s video, and options that may run analysis despite promising not to; one referenced path was missing, so evidence is incomplete.

The verdict is suspicious, not malicious: it indicates reported weaknesses or undisclosed behavior without establishing malicious intent. Consider the reports about video transfer, automatic account creation, local credential storage, and the ignored no-analysis option before installing. Skillspector scanned all 29 listed components but marked its analysis partial because a local reference could not be resolved; the AIG scan completed with no findings, which does not resolve that gap.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
Findings (22)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Database storage could support the declared system, but this chunk specifically implements general user-account management rather than storage or analysis of children's interactions. Its identity and token fields, account queries, and user-table migrations represent capabilities not explained by the description. No network transmission or token harvesting is shown. This assessment applies only to the supplied chunk; other files could implement the declared video-analysis functionality.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The supplied chunk implements account-backed API infrastructure rather than camera processing, interaction detection, or heatmap generation. Generic formatting, filesystem, and polling helpers could reasonably support the declared system and are not themselves mismatches. However, silent remote registration, credential-file access, persistent account management, and payment-skill promotion introduce substantive behaviors absent from the description. The request wrapper also adds authentication headers without restricting absolute URLs to a trusted service origin. This assessment is limited to the supplied chunk; it does not establish whether the advertised video-analysis functionality exists elsewhere.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The visible code does not substantiate the declared video-analysis purpose, and its only substantive method is an inactive, broken agent-chat wrapper. It does not actually execute the commented-out external command. Imported base classes and services are not supplied, so additional inherited behavior cannot be assessed; this finding is limited to the provided code.

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-social-interaction-analysis-analysis"
description: "Using fixed cameras in kindergartens or early-education centers, the system analyzes multi-person video to detect social-interaction behaviors among children, including approach (distance < 1 m), conversation (face-to-face with mouth movement), and cooperative play (collaborative play, chasing, etc.). | 通过幼儿园或早教中心的固定摄像头,分析多人视频,检测儿童之间的社交互动行为,包括接近(距离<1米)、对话(面对面且嘴部运动)、共同游戏(合作玩耍、追逐等)。系统定期生成社交互动热力图,为教师

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 documented no-analysis mode still executes video analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The --export-env-only flag promises not to analyze anything but is never checked after parsing; supplying an input still reaches analyze_video(). This bypasses the user's no-analysis choice and can initiate processing and remote submission of sensitive classroom video.

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 submitted to the analysis API without disclosure that the file will be uploaded.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The local-file branch reads the entire video into a multipart payload and passes it to the analysis API without an upload-specific notice or consent check. General references to API processing and guardian consent do not clearly establish authorization to transfer identifiable children's footage to an external processor, making this especially sensitive.

Content

No source excerpt is available for this finding.

The --export-env-only help text promises not to perform analysis, but the flag is ignored and execution can proceed to video analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The program parses --export-env-only but never branches on it, so a command supplied with video input still invokes analysis despite promising otherwise. In this skill's context, that can cause unintended processing and disclosure of children's video rather than a harmless configuration-only operation.

Content

No source excerpt is available for this finding.

save() is documented as creating a new record but can overwrite an existing record after an insertion failure (L221–L244).

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

Any insertion exception triggers updates by ID and then by username or real name, potentially replacing an existing user's identity and authentication-token fields instead of rejecting the creation. Name collisions or crafted records can therefore corrupt account associations; exploitation depends on caller access, but sensitive child-report accounts increase the consequences.

Content

No source excerpt is available for this finding.

ai_chat claims to invoke an analysis agent but never executes it and always returns None.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L26 explicitly says this method invokes the openclaw agent through subprocess, and L33-L34 promise an analysis result or session ID. However, command execution is commented out at L61-L66; the empty dictionary assigned at L68 causes result.stderr at L70 to raise AttributeError, which is caught and suppressed, leaving the method to return None without performing analysis.

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

L101 opens the supplied path in 'w' mode, immediately erasing existing contents even before anything is written. The generic method name does not disclose this destructive behavior, and there is no explanatory comment, docstring, confirmation, or operation-specific log; the exception handler only runs if opening fails.

Content

No source excerpt is available for this finding.

Video-analysis requests can silently trigger remote account registration and persistent identity storage, neither of which is disclosed by the manifest.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

An ordinary HTTP operation can automatically invoke a remote login endpoint with registration enabled, transmit the selected identity, and persist returned account information and tokens through the DAO. This account-provisioning side effect is not disclosed in the supplied skill description, creating a privacy and consent issue that is more consequential in a child-observation workflow; this code alone does not establish transmission of children's video.

Content

No source excerpt is available for this finding.

The HTTP wrapper can silently register a remote account and persist returned credentials.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The generic request wrapper hides remote account creation behind authentication recovery, using silent and register flags without an explicit provisioning decision. The surrounding implementation also saves returned credentials, making this an undisclosed persistent side effect rather than merely a network request; it overlaps the SDI-1 finding.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
84% confidence
Finding

This request sends the selected username as both openId and mobile to a remote service during automatic registration, without an explicit consent step in the shown flow. The identity selection can use API_SECRET_KEY, so some configurations may also copy a credential into identity fields; the actual endpoint configuration is not supplied, and this call does not show video transmission.

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
91% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested path does not exist, without any explanatory comment, docstring, prompt, or log. Because BaseEnum.init_subclass calls this loader at L053 and L058, importing the module can modify the filesystem without explicitly requesting a save.

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:

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