Back to skill

Security audit

Plant Growth Stage Recognition Skill | 植物生长阶段识别技能

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because scanners report undisclosed account, credential, and data-handling behavior that raises concerns beyond its stated plant-analysis purpose.

The verdict is suspicious, so review the reported account registration, credential storage and forwarding, and media-upload behavior before installing. The scanner completed most checks but marked analysis partial because two referenced paths were missing; those parts could not be assessed from the supplied results.

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 (26)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied chunk implements generic database access and a concrete sys_user model, rather than image processing, model inference, plant-stage classification, or agricultural output. Database persistence can be a supporting implementation detail, but the concrete functionality here manages user identities, personal information, and tokens, with no demonstrated connection to the declared plant-recognition purpose. These operations are available when the DAO is instantiated or its methods are called; the chunk does not demonstrate network transmission or token use. This assessment is limited to the supplied code, not any unseen components of the skill.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The visible implementation does not substantiate the declared plant-analysis purpose. Its only substantive method concerns an agent-chat interface, and that method is nonfunctional because result is a dictionary but is accessed through stderr and stdout attributes. The commented-out subprocess code does not execute and is not evidence of actual external-command execution. Imported base classes and API-service implementations were not supplied, so any plant-analysis behavior they might contain cannot be assessed.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

Generic file, JSON, date, polling, and HTTP helpers could legitimately support plant recognition and are not independently evidence of a mismatch. However, this chunk also implements an undeclared account provisioning, credential-management, and billing integration layer. The request wrapper attaches configured credentials even when given an absolute destination URL, rather than limiting authentication to the configured API host. Those behaviors materially exceed the declared agricultural image-analysis purpose. These capabilities are generally activated when the relevant methods are called; the chunk does not demonstrate that they have already executed. Because only a shared utility file is supplied, the absence of recognition logic here does not establish that the full skill lacks it.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

Generic AI-analysis configuration could support the declared skill, and the absence of classification logic in a configuration file alone is not a mismatch. However, DETAIL_EXPORT_URL explicitly targets BASE_URL_HEALTH with the path /health/order/api/getReportDetailExport?id=, introducing a health-report resource not explained by the plant-identification description. This chunk only configures endpoints; it does not itself make requests or export data.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
83% confidence
Finding

The visible code implements video submission and account-scoped analysis-history access rather than an explicitly plant-focused workflow. Video submission and file output could support the declared purpose, but history retrieval is an additional user-facing capability not disclosed in the description. The underlying skill implementation is not supplied, so it remains possible that the delegated analysis identifies plant growth stages; the absence of local computer-vision code alone is not evidence of a mismatch.

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-growth-stage-recognition-analysis"
description: "Accurately identifies key growth stages of plants from germination to fruiting based on computer vision and deep learning, provides structured data for precision agriculture decision support. | 植物生长阶段识别技能,基于计算机视觉与深度学习算法,精准识别植物从发芽到结果的全生命周期关键生长阶段,为精准农业提供科学决策支持"
version: "1.0.24"
license: "MIT-0"
---

# 🌱 Plant Growth Stage Recognition Skill | 植物生长阶段识别技能
> **智能分析中枢** · 图片/视频智能分析 · 结构化报告 · 历�

The HTTP wrapper forwards service credentials to unrestricted caller-supplied destinations.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The wrapper accepts arbitrary absolute HTTP or HTTPS URLs and automatically attaches API keys and authentication tokens without checking the destination. A caller-controlled URL can therefore disclose credentials to an attacker-controlled host, while plain HTTP also exposes them in transit.

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 option promises not to execute analysis, but the code still runs recognition.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The program parses --export-env-only but never checks it, so supplying that option with valid input still invokes recognition. A user expecting a configuration-only operation can unintentionally upload local media, trigger account initialization, or incur remote processing costs.

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 input files are passed to an analysis API without disclosure that their contents leave the local system.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The code reads the entire local input file, packages its contents as a multipart file payload, and passes it to the analysis service. No prompt, log, docstring, or comment in this file discloses the upload; analyzing a file does not inherently imply permission to transmit potentially private video or health-related content to a remote service.

Content

No source excerpt is available for this finding.

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

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The --export-env-only flag promises not to run analysis, but its parsed value is ignored and valid inputs proceed to analyze_video. This violates a meaningful no-processing expectation and can cause unintended media transmission, account activity, or processing charges.

Content

No source excerpt is available for this finding.

The service exposes general-purpose remote mutation and user-account lookup capabilities unrelated to plant growth-stage recognition.

Medium
Category
Excessive Agency
Confidence
90% confidence
Finding

The manifest describes image-based plant-stage identification and structured agricultural analysis, but this service exposes caller-selected remote add/edit/delete operations (L050–L069), HTTP PUT/DELETE forwarding (L077–L095), and username-based user lookup (L098–L100). Although network requests could support remote inference, these resource-management and account-discovery capabilities have no stated connection to the declared analysis purpose; this file establishes their availability, not that they are invoked.

Content

No source excerpt is available for this finding.

The skill includes user-profile and authentication-token persistence unrelated to plant growth-stage recognition.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The user model stores authentication tokens as ordinary string columns in a workspace SQLite database, with no explicit credential protection shown. Local persistence is not inherently malicious or unrelated to an API-backed skill, but storing reusable credentials alongside profile data increases exposure to other processes, skills, or backups that can read the database.

Content

No source excerpt is available for this finding.

The ai_chat docstring claims to invoke OpenClaw through subprocess, but the implementation never executes an agent.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L26 explicitly promises a subprocess invocation, but both the command arguments and subprocess.run call are commented out (L48–L66). Instead, L68 creates an empty dictionary and L70 accesses its nonexistent stderr attribute; the resulting exception is caught at L81–L82, so the method returns None without performing the documented analysis.

Content

No source excerpt is available for this finding.

FileUtil.open silently truncates existing files (L099–L101).

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The generic open helper uses mode 'w', which immediately erases an existing file's contents. No docstring, comment, confirmation, or user-facing log discloses this destructive behavior; exception handling does not warn about successful truncation.

Content

No source excerpt is available for this finding.

Plant-analysis requests implicitly register remote accounts and persist user credentials.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

An ordinary HTTP request can silently trigger remote registration with silent=1 and register=1, transmit an identity, and persist returned credentials. This introduces account creation and data-retention side effects not apparent from the plant-recognition description, although the code alone does not establish malicious intent.

Content

No source excerpt is available for this finding.

HTTP authentication silently registers remote users and persists returned credentials without disclosure (L550–L604).

Medium
Category
Output Handling
Confidence
94% confidence
Finding

Authentication setup creates remote users and saves returned credentials automatically rather than through a distinct user-approved operation. This is the same underlying behavior as SDI-1 and creates unexpected privacy and credential-retention risks, rather than a separate exploit.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
95% confidence
Finding

The registration POST transmits the selected username as both openId and mobile without an explicit consent step. More seriously, the caller preferentially selects API_SECRET_KEY as that username when configured, so this flow can send an actual secret in identity fields to the health-service endpoint, where it may be retained or logged.

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"):

Loading configuration silently creates directories and files, including during module import.

Low
Category
Output Handling
Confidence
94% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested configuration is absent, without a log, prompt, docstring, or explanatory comment disclosing this behavior. BaseEnum.init_subclass calls this loader during class definition (L053 and L058), so importing the module can modify the filesystem without an explicit save operation.

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)

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/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()