Back to skill

Security audit

Pet Oral Snapshot & Gum Redness Recognition | 宠物口腔抓拍与牙龈红肿识别

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this pet oral image analysis skill suspicious because scanners report undeclared report-history and identity handling, automatic account creation, and controls that may not prevent media analysis or upload.

Consider avoiding installation unless you are comfortable with media being sent to a remote service and with the reported account and history behavior. The scan was partial: one local path-like reference was missing, so evidence about that reference is incomplete. The scanners report security and privacy concerns, but do not establish malicious intent.

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 (23)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The documented fallback silently reuses a locally persisted default identity when upstream identity is absent, including for cloud report retrieval. On a shared installation, this can associate different users with the same report history; the evidence supports this identity-isolation risk, but does not independently establish the finding's broader claims about personal-information modification or token management.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

The analysis commands and report-formatting/output options align with the declared purpose or are supporting implementation details. Historical-report listing is a separate user-facing capability beyond analyzing supplied media and is not declared. The underlying module is not supplied, so actual API calls, storage access, and analysis behavior cannot be verified from these examples.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The main analysis workflow matches the declared purpose. However, retrieving an existing user's analysis list is an additional data-access capability, not merely an implementation detail of analyzing a supplied oral snapshot, and is absent from the description. Identity resolution alone would not establish a mismatch, but its use in this separate retrieval workflow does. The imported implementations are not supplied, so the assessment does not assume additional network behavior, authorization weaknesses, or diagnostic functionality.

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-pet-oral-snapshot-gum-redness-analysis"
description: "Triggers when a user provides an oral snapshot image/video of a pet (usually auto-captured during yawning, lip-licking or mouth-opening moments) for analysis; supports local uploads or network URLs to call server-side APIs for oral health recognition, evaluating gum color (pink / bright red / dark red) and tartar coverage area, outputting standardized oral health observations to help early discovery of periodontal disease (without diagnosing diseases). Application scenarios: pet cameras, smart pet products, pet health management platforms. | 当用户提供宠物口腔抓拍图�

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 remote-analysis workflow lacks a user-facing privacy warning about transmitting media and maintaining cloud report history.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

Although server-side processing and cloud history are mentioned, the automatic workflow does not require a user-facing explanation of the recipient, retained media or reports, or privacy implications before transmission. Pet-camera images can contain people and private household details, making silent upload and persistent history a meaningful privacy risk.

Content

No source excerpt is available for this finding.

--export-env-only promises not to analyze media, but does not prevent analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The parser accepts --export-env-only and promises no analysis, but execution never checks that flag. When media is supplied, a user expecting a non-analysis operation can instead trigger remote analysis and disclosure of the media.

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 service without an upload or privacy disclosure.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The method reads the complete local media file and passes it to the analysis service without a visible privacy acknowledgment or authorization gate. Remote analysis is an intended capability rather than covert exfiltration, but locally captured pet footage may expose people and household information if callers have not obtained informed upload authorization.

Content

No source excerpt is available for this finding.

The --export-env-only option promises not to run analysis, but the code ignores it and proceeds with analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The --export-env-only flag is parsed but never used to stop execution. Supplying it together with media still invokes analysis, violating the stated no-analysis boundary and potentially uploading private footage unexpectedly.

Content

No source excerpt is available for this finding.

save() is documented as creating a new record but can overwrite an existing user record.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

Any insertion exception triggers an update by ID and then potentially by username or realname, copying fields including identity and authentication tokens. A collision or attacker-influenced model can therefore overwrite an existing local user's credentials instead of failing creation; exploitation depends on control over the supplied model.

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
97% confidence
Finding

The generic open helper always uses mode 'w', which immediately erases an existing file's contents. No docstring, explanatory comment, confirmation, or action log discloses this destructive behavior; exception tracing only runs after an error.

Content

No source excerpt is available for this finding.

The API request helper silently provisions service accounts, beyond the declared oral-image analysis workflow.

Medium
Category
Excessive Agency
Confidence
90% confidence
Finding

The request helper automatically invokes a login endpoint with silent registration enabled when credentials are unavailable. Although authentication supports the server-side analysis workflow, implicit account creation introduces an undisclosed remote side effect and transmits an identity without a separate registration decision.

Content

No source excerpt is available for this finding.

An ordinary HTTP request can silently initiate remote account registration without disclosure.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

An ordinary request can trigger remote registration through the embedded authentication fallback without an explicit opt-in. This is the same underlying issue as the provisioning finding: an analysis request unexpectedly creates remote account state and discloses an identifier.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
91% confidence
Finding

External API traffic is expected for this skill, but this request sends the selected identity as both mobile and openId during automatic registration. The selection prioritizes ApiEnum.API_SECRET_KEY, so a configured secret can be duplicated into identity fields and potentially retained in account records or logs rather than handled solely as a credential.

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
92% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested path is absent, without a log, prompt, or explanatory comment/docstring. BaseEnum.init_subclass calls this loader during class creation (L044–L059), so merely importing the module can modify the filesystem without disclosing that side effect.

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:

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

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L26 explicitly promises an OpenClaw agent invocation, but both the command arguments and subprocess.run call are commented out (L48–L66). Instead, L68 assigns an empty dictionary to result; accessing result.stderr raises an exception that is caught, so the method returns without performing the documented analysis.

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/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