Back to skill

Security audit

Clawmobile Skill

Security checks for vulnerabilities and agentic risk

Overview

This Android automation skill is not proven malicious, but its default networking, token handling, and file-write behavior make it risky enough to review carefully before installing.

Install only in an isolated test environment or trusted local network. Replace the default token before any use, bind services to localhost, avoid exposing port 8765 remotely without TLS and authentication controls, do not store tokens in ~/.bashrc, and review any workflow or recording action before running it on accounts or apps containing sensitive data.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (4)

T09 · Insecure Skill Coding Practices

Error
Location
skill/api_server.py:374
Finding

Publicly Known Authentication Token Protects a Plaintext Device-Control Service

Content
View full analysis

Vulnerability Details

File Location: skill/api_server.py:374-387, skill/api_server.py:588; skill/client.py:40-49; config/settings.yaml:6-23; SKILL.md:675-689
Vulnerability Type: Hardcoded credential, insecure transport, and unsafe network exposure
Risk Level: High

Vulnerable Code

skill/api_server.py:374-387:

javascript
const AUTH_TOKEN = "clawmobile-secret-token-change-in-production";
const server_state = {
    running: false,
    current_task: null,
    start_time: Date.now()
};

function authenticate(req) {
    let authHeader = req.headers["authorization"];
    if (!authHeader || !authHeader.startsWith("Bearer ")) {
        return false;
    }
    let token = authHeader.substring(7);
    return token === AUTH_TOKEN;
}

skill/api_server.py:588:

javascript
server.listen(8765);

skill/client.py:40-49:

python
def __init__(self, host: str = "localhost", port: int = 8765, token: str = None):
    self.base_url = f"http://{host}:{port}"
    self.token = token or "clawmobile-secret-token-change-in-production"
    self.timeout = 300  # 5 minutes default

def _get_headers(self) -> Dict[str, str]:
    """Get request headers with authentication"""
    return {
        "Content-Type": "application/json",
        "Authorization": f"Bearer {self.token}"
    }

config/settings.yaml:6-23:

yaml
api:
  base_url: "http://localhost:8765"
  timeout: 30000
  max_retries: 3
  retry_delay: 1000

server:
  host: "0.0.0.0"
  port: 8765
  workers: 4
  debug: false

auth:
  token_env: "CLAWMOBILE_API_TOKEN"
  default_token: "clawmobile-secret-token-change-in-production"
  token_header: "Authorization"
  token_prefix: "Bearer"

Technical Analysis

The generated AutoX.js server authenticates requests with a static bearer token published in the repository. The Python client silently uses the same token wh ...[truncated 2131 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove all fallback and example secrets from executable code. Refuse to start the service if no explicit token is configured.
  2. Generate a cryptographically random token during setup, using at least 128 bits of entropy.
  3. Store the token in a protected secret store or environment variable rather than committed configuration.
  4. Bind the service to 127.0.0.1 by default. Require an explicit opt-in before listening on external interfaces.
  5. Use ADB port forwarding or another authenticated local tunnel for host-to-device communication.
  6. Require TLS for any remote network mode so bearer credentials and task contents are encrypted in transit.
  7. Compare authentication tokens with a constant-time comparison function.
  8. Add request rate limiting, failed-authentication throttling, request logging, and token rotation support.
  9. Clearly reject placeholder tokens such as your-autox-token-here and the current published default.

T09 · Insecure Skill Coding Practices

Error
Location
skill/client.py:179
Finding

Caller-Controlled Task ID Enables Path Traversal and Arbitrary JSON File Write

Content
View full analysis

Vulnerability Details

File Location: skill/client.py:179-198
Vulnerability Type: Path traversal and arbitrary file write
Risk Level: High

Vulnerable Code

python
def write_task(self, task: Dict[str, Any]) -> str:
    """
    Write task to file system

    Args:
        task: Task definition

    Returns:
        Task file path
    """
    import os
    import uuid

    # Ensure directories exist
    os.makedirs(self.tasks_dir, exist_ok=True)

    # Generate task ID if not provided
    if "id" not in task:
        task["id"] = f"task_{uuid.uuid4().hex[:6]}"

    # Write task file
    task_file = f"{self.tasks_dir}/{task['id']}.json"
    with open(task_file, 'w') as f:
        json.dump(task, f, indent=2)

    return task_file

Technical Analysis

WorkspaceFileManager.write_task() directly interpolates the caller-provided task["id"] into a filesystem path. The method generates a safe identifier only when the field is absent; it performs no validation when an identifier is supplied.

Path separators and parent-directory components are therefore accepted. For example, an identifier containing ../../ causes the constructed path to resolve outside self.tasks_dir. The project contains task-ID validation helpers, but this write path does not invoke them.

The file is opened in write mode, so an existing writable file at the resolved path is truncated and replaced with attacker-controlled JSON. No canonical-path containment check, exclusive creation, or symlink protection is applied.

Attack Path

  1. An attacker reaches an application path that accepts a custom task dictionary or otherwise influences its id field.
  2. The attacker supplies an identifier containing traversal components, such as ../../target.
  3. write_task() constructs a path equivalent to: <workspace>/tasks/../../target.json.
  4. The operating system normalizes the traversal ...[truncated 957 chars]
Remediation
View remediation

Remediation Suggestions

  1. Validate every supplied task ID before using it in a path. Use a strict allowlist such as ^[A-Za-z0-9_-]{1,64}$.
  2. Reject identifiers containing /, \, .., null bytes, or platform-specific path separators.
  3. Construct paths with pathlib.Path, resolve the candidate path, and verify that it is a child of the resolved task directory.
  4. Do not rely solely on the existing validate_task_id() helper unless its format is updated to support all legitimately generated IDs consistently.
  5. Use exclusive file creation where overwriting is not required.
  6. Apply restrictive file permissions and protect against symlink-based redirection.
  7. Add tests covering absolute paths, parent traversal, nested separators, encoded separators, and symlink targets.

T09 · Insecure Skill Coding Practices

Error
Location
skill/api_server.py:278
Finding

Membership Privileges Can Be Forged Using Locally Generated Redeem Codes

Content
View full analysis

Vulnerability Details

File Location: skill/api_server.py:278-313; skill/models.py:346-418
Vulnerability Type: Broken authorization and forgeable entitlement validation
Risk Level: High

Vulnerable Code

skill/api_server.py:278-313:

python
@staticmethod
def activate_redeem_code(code: str, user_id: str) -> Dict[str, Any]:
    """
    Activate membership using redeem code
    POST /api/v1/membership/redeem
    """
    if not code or not user_id:
        return {
            "success": False,
            "error": "INVALID_CODE",
            "message": "兑换码格式无效"
        }

    # Validate code format
    if not MembershipAPI._validate_code_format(code):
        return {
            "success": False,
            "error": "INVALID_CODE",
            "message": "兑换码格式无效"
        }

    # Parse tier from code
    parts = code.split('-')
    tier_str = parts[0].lower()
    tier = tier_str if tier_str in ["vip", "svip"] else "vip"

    return {
        "success": True,
        "message": "兑换成功!",
        "tier": tier,
        "duration_days": 30,
        "expires_at": "2026-04-30T15:00:00Z"
    }

@staticmethod
def _validate_code_format(code: str) -> bool:
    """Validate redeem code format"""
    import re
    pattern = r'^(VIP|SVIP)-\d{4}-[A-Z0-9]{8}$'
    return bool(re.match(pattern, code))

skill/models.py:346-418:

python
@staticmethod
def activate_membership(code: str, user_id: str) -> MembershipInfo:
    """
    Activate membership using redeem code

    Returns:
        MembershipInfo after activation
    """
    # Parse code
    parts = code.split('-')
    tier_str = parts[0].lower()

    if tier_str == "vip":
        tier = MembershipTier.VIP
        days = 30
    elif tier_str == "svip":
        tier = MembershipTier.SVIP
        days = 90
    else:
        raise ValueError("Invalid tier in code")


...[truncated 3054 chars]
Remediation
View remediation

Remediation Suggestions

  1. Treat redeem-code verification as a server-side authorization operation.
  2. Store issued codes in a trusted database with tier, expiration, redemption status, intended user, and issuer metadata.
  3. Alternatively, use digitally signed redeem tokens and verify signatures with a trusted public key.
  4. Enforce single-use redemption atomically to prevent replay and race conditions.
  5. Validate user identity before assigning an entitlement.
  6. Never derive authoritative privileges solely from a caller-controlled string or client-side membership object.
  7. Ensure all activation entry points call the same centralized verifier.
  8. Record security audit events for successful and failed redemption attempts.
  9. Add tests for fabricated codes, expired codes, reused codes, altered tiers, invalid users, and concurrent redemption.

T08 · Insecure Dependencies

Warning
Location
scripts/setup.sh:76
Finding

Setup Installs Unlocked Dependencies and Documentation Downloads an Unverified Skill Archive

Content
View full analysis

Vulnerability Details

File Location: scripts/setup.sh:76-91; requirements.txt:1-18; SKILL.md:305-311
Vulnerability Type: Unverified software retrieval and insufficient dependency pinning
Risk Level: Medium

Vulnerable Code

scripts/setup.sh:76-91:

bash
check_dependencies() {
    print_info "Checking Python dependencies..."

    if [ ! -f "$SKILL_ROOT/requirements.txt" ]; then
        print_warning "requirements.txt not found"
    else
        print_info "Installing dependencies..."
        pip3 install -q -r "$SKILL_ROOT/requirements.txt" || {
            print_error "Failed to install dependencies"
            exit 1
        }
        print_info "Dependencies installed ✓"
    fi
}

requirements.txt:1-18:

text
# ClawMobile Skill Python Dependencies

# HTTP Requests
requests>=2.31.0

# YAML Configuration
PyYAML>=6.0

# Optional: Testing dependencies
pytest>=7.0.0
pytest-cov>=4.0.0

# Optional: Code quality
flake8>=6.0.0
black>=23.0.0

# Optional: Type checking
mypy>=1.0.0

SKILL.md:305-311:

bash
# 下载 Skill 包
wget https://github.com/clawmobile/clawmobile/archive/v1.0.0.zip

# 解压到技能目录
unzip v1.0.0.zip -d ~/.openclaw/skills/
mv ~/.openclaw/skills/clawmobile-1.0.0 ~/.openclaw/skills/clawmobile

Technical Analysis

The setup script runs pip3 install against dependency constraints that specify only minimum versions. Future versions satisfying these ranges can be installed without having been reviewed with the Skill. No lock file or package hashes are supplied.

The runtime installation also includes testing, formatting, linting, and type-checking packages described as optional. Installing these packages in normal setup unnecessarily increases the number of packages and transitive dependencies trusted by the environment.

The manual installation instructions download and unpack a complete external Skill archive with ...[truncated 2041 chars]

Remediation
View remediation

Remediation Suggestions

  1. Create a reproducible lock file containing exact dependency versions.
  2. Use hash verification, such as pip's --require-hashes, for all packages and transitive dependencies.
  3. Separate runtime dependencies from development and testing dependencies.
  4. Install only requests and any genuinely required runtime libraries during normal Skill setup.
  5. Review and update locked dependencies through a controlled dependency-management process.
  6. Publish checksums or cryptographic signatures for downloadable Skill archives.
  7. Verify the archive before extracting it into the active skills directory.
  8. Download to a staging directory, validate expected paths and contents, and only then perform installation.
  9. Synchronize the documented archive version with the Skill's declared version.
  10. Prefer the package manager installation workflow where it provides authenticated manifests and integrity validation.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
Findings (55)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · CHANGELOG.md (reported line 64)May include surrounding context.

md
- ✅ `GET /workflows/{id}` - Get workflow details
- ✅ `POST /workflows` - Create workflow
- ✅ `PUT /workflows/{id}` - Update workflow
- ✅ `DELETE /workflows/{id}` - Delete workflow
- ✅ `POST /recording/start` - Start recording
- ✅ `POST /recording/pause` - Pause recording
- ✅ `POST /recording/resume` - Resume recording

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The description presents a full-featured Android automation/RPA toolkit. However, the provided code chunk is only a test script for that toolkit. Its purpose is to verify imports, configuration files, model instantiation, membership code validation, metadata files, and optional API availability. While these tests reference components consistent with the broader product domain, the actual code shown does not perform the declared end-user functionality. Therefore, for this supplied code chunk, the declared description does not accurately represent what the code actually does.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared description presents a full-featured Android automation/RPA toolkit with runtime capabilities such as workflow management, recording, AI intervention, membership handling, and mobile interaction. However, the supplied code chunk does not implement those functions. It is only a validation utility for checking repository structure, Python files, config syntax, dependencies, environment variables, ADB presence, and API connectivity. While some checks reference components consistent with the declared ecosystem (ADB, API server, Python modules), the actual behavior of this code is limited to setup validation and readiness checks. That is a materially different primary purpose from the declared end-user functionality, so this chunk does not accurately represent the declared description.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The description presents a full-featured Android automation toolkit with workflow management, task recording, AI intelligent intervention, membership system, HTTP API communication, and broad RPA/mobile automation capabilities. The supplied code only partially matches that description. It does include HTTP API communication and some membership-related logic, but the automation side is a simplified mock/MVP server that simulates task execution rather than carrying out real Android automation. The status endpoint explicitly marks recording and AI intervention as false, which directly conflicts with the declared feature set. There is no real recording/playback engine, no meaningful AI behavior, and only minimal workflow/state tracking. Therefore, the description materially overstates the implemented functionality in this code chunk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The supplied code does fit part of the declared Android automation/RPA purpose: it executes mobile interaction tasks like click, input, and swipe through an AutoX.js client and workspace files. However, the declaration presents a much broader, 'complete toolkit' including workflow management, recording, AI intervention, membership features, and HTTP API communication. None of those broader capabilities are implemented in this code chunk itself. The chunk’s actual primary purpose is a limited execution wrapper for basic automation actions, so the description materially overstates the represented functionality for this specific code.

Content

No source excerpt is available for this finding.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
87% confidence
Finding

The skill exposes destructive workflow operations such as DELETE /workflows/{id} and is designed for broad natural-language triggering, yet the documentation does not describe confirmation requirements, authorization boundaries, or safeguards around destructive actions. In an automation tool, unintended or manipulated deletion of workflows can disrupt operations and remove business-critical automations.

Content

Scanner excerpt · SKILL.md (reported line 160)May include surrounding context.

md
- `GET /workflows/{id}` - 获取工作流详情
- `POST /workflows` - 创建工作流
- `PUT /workflows/{id}` - 更新工作流
- `DELETE /workflows/{id}` - 删除工作流
- `POST /api/v1/workflows/{id}/validate` - 验证工作流

### 2. 录制功能 ⏺️

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill promotes recording screenshots, UI trees, and unattended automation, but it does not prominently warn users that these features may capture sensitive content such as messages, credentials, personal data, or app state. Because the skill targets full device automation, lack of disclosure meaningfully increases privacy and security risk.

Content

No source excerpt is available for this finding.

YARA rule 'backdoor_persistence': Backdoor persistence with malicious payloads (shell commands, SSH key injection, hidden root users) [malware]

High
Category
YARA Match
Confidence
83% confidence
Finding

Appending exported variables to ~/.bashrc persists the API URL and API token in plain text, which can expose credentials to other local users, backups, shell history workflows, or accidental disclosure. Although presented as convenience setup, persisting authentication material in a startup file is a risky secret-management practice.

Content

Scanner excerpt · SKILL.md (reported line 402)May include surrounding context.

ion: 3600 capture_screenshots: true capture_ui_tree: true

日志配置

logging: level: "INFO" file: "clawmobile.log" max_size: "10MB" backup_count: 5 EOF

text

**方法 B:使用环境变量**

```bash
# 设置环境变量
export CLAWMOBILE_API_URL="http://localhost:8765"
export CLAWMOBILE_API_TOKEN="your-autox-token-here"
export CLAWMOBILE_TIMEOUT="30000"

# 或添加到 ~/.bashrc
echo 'export CLAWMOBILE_API_URL="http://localhost:8765"' >> ~/.bashrc
echo 'export CLAWMOBILE_API_TOKEN="your-autox-token-here"' >> ~/.bashrc
source ~/.bashrc

4. 验证连接

bash
# 测试基本连接
curl http://localhost:8765/api/v1/health

# 预期响应:
# {"success": true, "health": {"status": "healthy", ...}}

# 使用 Python 客户端测试
python3 -c "
from skill.client import ClawMobileClient
client = ClawMobileClient()
health = client.get_health()
print(f'Health: {health}')
"

基本用法

使用 Python 客户端

python
from skill.client import ClawM

Instruction Override

High
Category
Prompt Injection
Confidence
70% confidence
Finding

This pattern attempts to override system instructions or ignore safety constraints. Without LLM analysis, manual review is recommended.

Content

Scanner excerpt · config/schema.json (reported line 55)May include surrounding context.

json
},
        "debug": {
          "type": "boolean",
          "description": "Enable debug mode"
        }
      },
      "required": ["host", "port"]

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/setup.sh (reported line 113)May include surrounding context.

sh
setup_env() {
    print_info "Setting up environment variables..."

    ENV_FILE="$SKILL_ROOT/.env"

    if [ -f "$ENV_FILE" ]; then
        print_info ".env file already exists"

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/setup.sh (reported line 197)May include surrounding context.

sh
setup_env() {
    print_info "Setting up environment variables..."

    ENV_FILE="$SKILL_ROOT/.env"

    if [ -f "$ENV_FILE" ]; then
        print_info ".env file already exists"

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/setup.sh (reported line 116)May include surrounding context.

sh
ENV_FILE="$SKILL_ROOT/.env"

    if [ -f "$ENV_FILE" ]; then
        print_info ".env file already exists"
        read -p "Update .env file? (y/n) " -n 1 -r
        echo
        if [[ ! $REPLY =~ ^[Yy]$ ]]; then

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/setup.sh (reported line 117)May include surrounding context.

sh
ENV_FILE="$SKILL_ROOT/.env"

    if [ -f "$ENV_FILE" ]; then
        print_info ".env file already exists"
        read -p "Update .env file? (y/n) " -n 1 -r
        echo
        if [[ ! $REPLY =~ ^[Yy]$ ]]; then

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/setup.sh (reported line 137)May include surrounding context.

sh
ENV_FILE="$SKILL_ROOT/.env"

    if [ -f "$ENV_FILE" ]; then
        print_info ".env file already exists"
        read -p "Update .env file? (y/n) " -n 1 -r
        echo
        if [[ ! $REPLY =~ ^[Yy]$ ]]; then

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/setup.sh (reported line 138)May include surrounding context.

sh
ENV_FILE="$SKILL_ROOT/.env"

    if [ -f "$ENV_FILE" ]; then
        print_info ".env file already exists"
        read -p "Update .env file? (y/n) " -n 1 -r
        echo
        if [[ ! $REPLY =~ ^[Yy]$ ]]; then

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/setup.sh (reported line 196)May include surrounding context.

sh
ENV_FILE="$SKILL_ROOT/.env"

    if [ -f "$ENV_FILE" ]; then
        print_info ".env file already exists"
        read -p "Update .env file? (y/n) " -n 1 -r
        echo
        if [[ ! $REPLY =~ ^[Yy]$ ]]; then

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · skill/api_server.py (reported line 507)May include surrounding context.

python
}
    }

    // Reset state
    server_state.running = false;
    server_state.current_task = null;

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The changelog documents capabilities for remote workflow execution, UI recording, screenshot capture, UI tree capture, and destructive delete operations, but provides no warning about privacy-sensitive data capture, operator consent, or destructive effects. In a mobile automation/RPA skill, these features can expose sensitive on-screen data or enable harmful actions if users are not clearly warned and guided, making the omission a real security concern even though it appears in documentation rather than executable code.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This markdown file presents complete device control, workflow execution, recording, and AI-driven recovery as core features, but does not disclose that these actions may interact with apps, modify device state, or affect user data/privacy. For a skill that automates Android devices through an HTTP server, the description should warn users about the operational and privacy impact before use.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The README advertises operation recording and later shows how to start recording for a target app, but it does not warn that recordings may capture sensitive UI content, credentials, personal data, or other private information. This is a user-facing omission in markdown for a privacy-relevant behavior.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · README.md (reported line 124)May include surrounding context.

健康检查

bash
curl http://localhost:8765/api/v1/health

列出工作流

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding

The skill advertises and demonstrates network access, shell commands, and file operations, but it does not clearly declare a restrictive tool scope for those capabilities. That creates an authorization and review gap: operators may enable the skill without realizing it can write files, enumerate directories, invoke adb/curl, and persist configuration.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

Natural-language instructions, triggers, examples, and operational guidance are presented in Chinese throughout the skill, effectively forcing a specific language for users. The document does not offer an opt-in language choice or explain that the skill is intentionally limited to a Chinese-speaking or region-specific audience.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill requires an API token and describes HTTP API communication, but it does not clearly warn about transmitting tokens and automation data over HTTP, including the insecure localhost example and possible port-forwarded usage. Users may copy this pattern into non-local deployments, exposing credentials or captured automation data without transport protection.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The trigger list includes broad terms like RPA, batch execution, AI intervention, and automation-related phrases that could activate the skill in contexts the user did not intend. In a skill capable of device control, unattended execution, recording, and network access, accidental invocation increases the chance of unintended actions on connected Android devices.

Content

No source excerpt is available for this finding.