Back to skill

Security audit

Greenhouse Climate Plant Feedback Analysis | 温室环境与植物状态联动调控

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because its reported account, credential, and report-management features do not clearly match its greenhouse-analysis description, and scanners flag unintended external and persistent side effects.

Treat this as a review-needed skill and consider avoiding installation until the account and credential handling, external media transfers, and export-only behavior are clearly explained and addressed. Skillspector inspected all 29 listed components, but its analysis is marked partial because a local reference was missing; the separate AIG scan completed with no findings, which does not resolve the concerns raised by the other scanner.

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 (27)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied chunk is primarily user-account persistence infrastructure. Its concrete model stores names, email, birthday, sex, age, token, and open_token, and its DAO creates or alters the sys_user table and supports reading, updating, and deleting records. Although database storage can support a greenhouse application, this code contains no greenhouse-specific data or processing, and the declared purpose does not describe identity or token management. No camera access, sensor ingestion, AI inference, or actuator-command generation appears in the supplied chunk. This assessment is limited to the provided code; it does not establish malicious intent or token exfiltration.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied code is an API-backed video-analysis and report-management wrapper. It contains no visible processing of environmental sensor readings, greenhouse-specific morphology interpretation, or generation of irrigation, shade, fan, wet-curtain, or heater commands. Although the delegated API could perform analysis not shown here, its greenhouse-specific behavior cannot be established from this chunk. Local video upload and polling could support the declared analysis, but the implemented reporting workflow and report-history access materially differ from the stated purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

Basic formatting, polling, and HTTP utilities could support a greenhouse application and are not themselves mismatches. However, this chunk additionally implements an undeclared identity, authentication, registration, and billing integration. Its request helper accepts absolute URLs and adds configured credentials without visibly limiting destinations. No camera analysis, sensor processing, AI climate decision-making, or actuator-control logic appears in the supplied chunk. This assessment is limited to the shown code; it does not establish whether greenhouse functionality exists elsewhere.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The supplied code does not implement the declared primary purpose. Its only substantive method prepares generic agent-chat session metadata, while the intended external command invocation is disabled and the remaining result handling fails. Imported base classes and services may contain additional functionality, but their behavior cannot be established from this chunk. No active subprocess execution or malicious resource access is demonstrated.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

Generic AI-analysis configuration could support the declared purpose, and missing greenhouse logic in a configuration-only chunk is not itself evidence of a mismatch. However, DETAIL_EXPORT_URL explicitly targets BASE_URL_HEALTH plus /health/order/api/getReportDetailExport?id=, indicating a healthcare/order-report resource inconsistent with the declared greenhouse function. This chunk only configures that endpoint; it does not demonstrate an actual request or data export.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

The supplied code implements video submission and analysis-history retrieval. It contains no visible environmental-sensor integration, plant-morphology processing, continuous camera monitoring, or generation of greenhouse control commands. Analysis is delegated to an unavailable implementation, so greenhouse-specific processing inside that dependency cannot be ruled out. Nevertheless, the visible functionality does not substantiate the declared primary purpose, and its history-retrieval capability is undeclared.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
86% confidence
Finding

Generic HTTP helpers could support the declared greenhouse functionality and are not inherently a mismatch. However, the supplied chunk additionally exposes storage-download and user-directory operations that the description does not disclose or connect to that purpose. It contains no image analysis, sensor processing, AI decision logic, or climate-control command generation. This assessment is limited to the supplied chunk; it does not establish malicious behavior or rule out greenhouse functionality elsewhere.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
81% confidence
Finding

Greenhouse terminology appears in help and status messages, but the executable workflow only updates a pet-type constant and delegates the media input to skill.get_output_analysis. There is no visible sensor-data ingestion, continuous camera processing, sensor/image fusion, or climate-command generation. The imported analysis implementation is not supplied, so its downstream capabilities cannot be established; this assessment concerns the substantial gap between the declared integrated workflow and the visible implementation. Identity resolution, history retrieval, and output-file writing are not independently treated as mismatches.

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-greenhouse-climate-plant-feedback-analysis"
description: "Using fixed cameras in a smart greenhouse to analyze plant morphology in real time (e.g., leaf wilting angle, stem uprightness, leaf color changes) combined with environmental sensors (light intensity, temperature, humidity, soil moisture), an AI decision model outputs climate control commands including irrigation (pump/solenoid valve), shade-net opening, fan/wet-curtain on-off, heater on-off, etc. | 通过智能温室中的固定摄像头实时分析植物的形态(如叶片萎蔫角度、茎秆挺直度、叶色变化)以及结合环境传感器(光照强度、温度、�

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 an explicit privacy warning about sending local images or videos to an external API.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow automatically sends selected local media for external analysis without a clear pre-upload privacy notice or approval boundary. Although API use is mentioned, greenhouse footage may reveal workers, facility layouts, or commercially sensitive crop information that users do not expect to leave the environment.

Content

No source excerpt is available for this finding.

The documented export-only mode still executes 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/history execution. With an input supplied, a supposedly non-executing invocation can upload media and trigger remote processing or charges.

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 run analysis, but the code still invokes analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The --export-env-only argument is parsed but never consulted, so supplying it with media still reaches the normal analysis call. This violates the promised no-analysis boundary and can cause unintended external disclosure, processing costs, or account-related side effects.

Content

No source excerpt is available for this finding.

Configuration loading silently creates directories and writes missing configuration files.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested path does not exist, without a prompt, log, docstring, or explanatory comment disclosing this side effect. BaseEnum.init_subclass invokes this loader during class creation (L044–L059), so merely importing the module can modify the filesystem without an explicit save operation.

Content

No source excerpt is available for this finding.

The skill includes user-profile and credential database management unrelated to greenhouse analysis and climate control.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

L448-L473 define a user-account model containing personal information, token, and open_token fields, and expose inherited CRUD operations through UserDao. Database initialization also automatically creates tables and alters sys_user fields (L180-L214). Managing user identities and credentials is not a direct requirement of the manifest's plant morphology, environmental sensing, or climate-command functionality.

Content

No source excerpt is available for this finding.

ai_chat documents an AI-agent invocation but never executes one or returns an analysis result.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L26 explicitly states that this method invokes the OpenClaw agent through subprocess, but the command and subprocess invocation are commented out at L48–L66. Instead, L68 creates an empty dictionary and L70 accesses its nonexistent stderr attribute; the resulting exception is caught at L81–L82, leaving the method to return None without performing the documented AI analysis.

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
97% confidence
Finding

The generically named open helper uses mode 'w', which immediately erases an existing file's contents. There is no docstring, explanatory comment, confirmation, or operation-specific log warning about truncation; exception tracing does not disclose a successful overwrite.

Content

No source excerpt is available for this finding.

Ordinary API requests can silently register a remote user and persist account credentials, beyond the stated greenhouse-analysis purpose.

Medium
Category
Excessive Agency
Confidence
92% confidence
Finding

An ordinary request can invoke a separate login endpoint with silent registration enabled, transmit an identity, and persist returned credentials without an explicit enrollment step. This introduces undisclosed remote account creation and credential retention beyond greenhouse analysis, although malicious intent and server-side misuse are not established.

Content

No source excerpt is available for this finding.

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

Medium
Category
Output Handling
Confidence
93% confidence
Finding

The request helper hides account registration and local credential persistence inside a routine HTTP operation. These security-relevant side effects may occur without the operator understanding or approving them, creating the same consent and credential-lifecycle issue identified by SDI-2.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

External communication alone is not a vulnerability, but this call receives current__user_name, which preferentially takes ApiEnum.API_SECRET_KEY when configured. If tokens are missing, that secret can therefore be copied into both openId and mobile in a login/registration payload, unnecessarily exposing it to identity-processing and logging systems; the endpoint's actual trust level is not provided.

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"):

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)

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: