Back to skill

Security audit

Plant Wilting Monitoring Skill | 植物枯萎监测技能

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this plant-wilting skill suspicious because scanners report account, credential, and data-handling behavior beyond its stated purpose, along with weakly scoped triggers and execution controls.

Consider the suspicious verdict before installing. The reported account creation, credential handling, possible video uploads, and export-only behavior deserve particular attention. Skillspector reports partial coverage because two referenced local paths were missing, so the available evidence is incomplete; the clean result from the separate AIG scan does not resolve these concerns.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (26)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied chunk contains no hyperspectral processing, computer-vision analysis, wilting detection, or early-warning logic. Database persistence alone could be a supporting implementation detail, but this code specifically implements user identity and token management, a materially different capability not described in the declared purpose. The assessment is limited to this chunk; it does not establish that plant-monitoring functionality is absent elsewhere or that stored data is transmitted externally.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

Generic file, JSON, date, and request helpers could legitimately support a plant-monitoring skill; their presence alone is not a mismatch. However, this chunk additionally implements an undeclared account-registration, credential-management, and billing workflow. Its request wrapper can transmit configured credentials to arbitrary supplied HTTP(S) destinations without a visible origin check. These are material capabilities beyond the declared description. The supplied chunk contains no plant-analysis implementation, although other unseen files could provide that functionality. Most behavior occurs when the utility methods are invoked, rather than simply upon importing the module.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The supplied chunk does not implement the declared plant-wilting monitoring purpose. Its visible behavior is generic scaffolding and an incomplete chat wrapper. No external agent command actually executes because that code is commented out. Implementations of the imported base classes and services are not supplied, so their behavior cannot be assessed; this 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
92% confidence
Finding

The supplied code presents itself as a video-analysis tool and contains no visible hyperspectral handling, plant-specific analysis, early-wilting detection, or irrigation/disease warning logic. It also exposes an analysis-history listing capability tied to a resolved user identifier. The imported skill implementation is not provided, so it could contain domain-specific processing; nevertheless, the behavior demonstrated by this chunk does not substantiate the stated specialized purpose and includes an undeclared history-retrieval capability.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
82% confidence
Finding

DETAIL_EXPORT_URL combines BASE_URL_HEALTH with /health/order/api/getReportDetailExport?id=, indicating a health-order reporting resource not explained by the declared purpose. Generic AI-analysis configuration could support the stated workflow and is not independently a mismatch. The absence of plant-analysis logic in this configuration-only chunk is also not sufficient evidence by itself. The mismatch finding rests on the explicitly configured health-report resource; its actual use is not demonstrated here.

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: "plant-wilting-monitoring-analysis"
description: "Early monitoring of plant wilting based on hyperspectral imaging and computer vision, captures early wilting signs before visible symptoms, provides early warning for precision irrigation and disease control. | 植物枯萎监测技能,基于高光谱成像与计算机视觉,在肉眼可见症状前捕捉早期枯萎迹象,为精准灌溉和病害防控提供早期预警"
version: "1.0.25"
license: "MIT-0"
---

# 🥀 Plant Wilting Monitoring Skill | 植物枯萎监测技能
> **智能分析中枢** · 图片/视频智能分析 · 结构化报告 · 历史报告云端查询

Caller-supplied absolute URLs receive shared service credentials without a destination restriction.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The helper accepts arbitrary absolute HTTP or HTTPS URLs and then adds shared X-Access-Token, X-Api-Key, and Authorization credentials without checking the destination. A caller-controlled destination can therefore receive service secrets, and accepting plain HTTP also permits unencrypted credential transmission.

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.

Generic history-report phrases are listed as automatic triggers without requiring a plant-wilting context.

Medium
Category
Trigger Abuse
Confidence
94% confidence
Finding

The examples designate generic phrases such as 'historical reports' as automatic triggers, while the workflow mandates a cloud query once triggered. In a multi-skill agent, an unrelated report request could therefore cause unintended identity-associated network access and disclosure of plant-monitoring history.

Content

No source excerpt is available for this finding.

The documented export-only mode still executes plant-image analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The parser promises that --export-env-only will not execute analysis, but the parsed option is never checked. With valid input, execution still initializes identity and invokes analysis, potentially uploading media or incurring service charges despite the caller selecting a non-executing mode.

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 of the upload.

Medium
Category
Output Handling
Confidence
87% confidence
Finding

The skill reads the entire local video into a file payload and passes it to the analysis API. Although a comment documents reading the file, nothing in this flow explains that its contents will be transmitted, which could expose private video data unexpectedly.

Content

No source excerpt is available for this finding.

The --export-env-only option promises not to execute analysis, but the code still runs analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The --export-env-only argument is accepted but never used to control execution. Supplying valid media input still runs identity initialization and video analysis, violating the promised no-analysis boundary and potentially causing unintended uploads or charges.

Content

No source excerpt is available for this finding.

The skill includes user-profile and credential storage unrelated to plant-wilting analysis.

Medium
Category
Excessive Agency
Confidence
91% confidence
Finding

The user model stores account tokens and personal-profile fields in ordinary SQLite columns, with no application-level secret protection or restrictive file-permission enforcement visible in this DAO. If populated, access to the database or its backups exposes reusable credentials; this sensitive persistence expands the risk beyond the advertised plant-analysis functionality, although the schema alone does not prove data collection.

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
96% confidence
Finding

At L101, opening a path in 'w' mode immediately erases any existing contents. The generically named helper has no docstring, explanatory comment, confirmation, or operation log warning callers about this behavior; exception tracing does not disclose successful truncation.

Content

No source excerpt is available for this finding.

The request helper silently registers or logs in a remote user and persists account credentials, beyond the described plant-analysis workflow.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

An ordinary request can implicitly invoke phoneLogin with registration enabled, transmit the selected identity, and save returned credentials without a separate authorization or consent step. Remote account provisioning might support a hosted analysis service, but hiding it inside a generic request helper creates unexpected privacy and persistent-account side effects not disclosed by the plant-monitoring description.

Content

No source excerpt is available for this finding.

HTTP requests can trigger undisclosed remote account registration.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The request helper sets both silent and register to 1 when obtaining a remote account, so a routine HTTP operation can create persistent remote state without an explicit provisioning decision. This is an unexpected side effect for a plant-analysis skill, even if the remote service legitimately requires an account.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
94% confidence
Finding

This POST transmits the selected identity as both openId and mobile during implicit login or registration, without a visible consent gate. The calling logic can select API_SECRET_KEY as that identity, potentially placing a credential in identity fields and downstream account records; the configured endpoint's ownership and transport security are not shown.

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 a configuration file, including during module import.

Low
Category
Output Handling
Confidence
93% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested path is absent, without a comment, docstring, log, or confirmation disclosing this side effect. BaseEnum.init_subclass invokes this loader during class definition (L052–L053), so simply importing the module can modify the filesystem rather than only read 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:

ai_chat documents an OpenClaw subprocess call and analysis output, but never invokes the agent and always returns None.

Low
Category
Excessive Agency
Confidence
100% 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, the subprocess invocation is commented out at L61–L66, and accessing stderr on the empty dictionary created at L68 raises an AttributeError that is caught and suppressed, causing the method to return None without performing analysis.

Content

No source excerpt is available for this finding.