Back to skill

Security audit

华米(Zepp/原小米)运动刷步数

Security checks for vulnerabilities and agentic risk

Overview

This skill modifies Zepp/Xiaomi step counts, but it handles account passwords unsafely and includes risky credential and automation guidance.

Review carefully before installing. Do not enter Zepp/Xiaomi passwords into the listed third-party testing websites. Avoid storing real account passwords in this skill's config.json unless you accept the local exposure risk, and do not enable scheduled runs unless you have reviewed the exact cron job and know how to remove it. Prefer a version that prompts for secrets per run or uses a secure secret store and removes fake source-IP headers.

Vulnerability Patterns
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (4)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:52
Finding

Plaintext Storage of Reusable Account Credentials

Content
View full analysis
Remediation
View remediation

other

Error
Location
SKILL.md:36
Finding

Unsafe Recommendation to Submit Account Credentials to Unrelated Third-Party Websites

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/huami.py:12
Finding

Unnecessary Source-IP Header Spoofing During Authentication

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/huami.py:151
Finding

Reusable Login Token Exposed in a URL Query String

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (18)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill tells the agent to ask for a Zepp/Xiaomi account password and store it in config.json in plaintext, without an upfront warning before collection. This is dangerous because credentials are highly sensitive, can be exposed through logs, backups, or filesystem access, and are being collected for a service-manipulation workflow rather than a clearly safety-reviewed authentication flow.

Content

No source excerpt is available for this finding.

Ssd 3

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill explicitly instructs the agent to solicit user credentials and persist them locally for later use. This creates a clear secret-handling vulnerability: the agent becomes a collector and custodian of third-party account passwords, which greatly increases the risk of credential theft, unintended reuse, and compromise of external accounts.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · scripts/huami.py (reported line 38)May include surrounding context.

python
use_fake = should_use_fake_ip() if use_fake_ip is None else use_fake_ip
    fake_addr = fake_ip() if use_fake else None

    # 1. Get access token
    url1 = "https://api-user.zepp.com/v2/registrations/tokens"
    login_headers = {
        "content-type": "application/x-www-form-urlencoded; charset=UTF-8",

Missing User Warnings

High
Category
Not specified by scanner
Confidence
94% confidence
Finding

The function performs live credentialed login, token acquisition, and remote data submission with user-supplied account credentials but contains no built-in confirmation, warning, or consent gate. In a skill explicitly meant to falsify fitness records, this increases the chance of users being tricked into handing over credentials and authorizing account actions they may not fully understand.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding

The skill performs environment inspection and networked actions but does not declare any explicit tool scope or permissions boundary. That omission makes the skill harder to sandbox and review, increasing the chance an agent will grant broader capabilities than necessary when handling credentials and remote API calls.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The trigger phrases are broad enough to match ordinary conversation about step counts, fitness bands, or Zepp/Xiaomi apps. This can cause the skill to activate unexpectedly and start credential collection or account-modification workflows without sufficiently specific user intent.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

The invocation guidance repeats ambiguous triggers without confirmation gates or exclusion rules, so a normal request mentioning step changes could launch a sensitive workflow. In this skill, that matters more because activation can lead directly to soliciting account credentials and performing remote account actions.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill directs the agent to create cron jobs for repeated execution, which introduces persistence beyond the immediate user request. Persistence changes the trust model because it can continue modifying data and using stored credentials after the session ends, increasing operational and security risk.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
80% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 143)May include surrounding context.

md
⚠️ 密码明文存储在 config.json 中:
- 仅在本地环境使用
- 不要将 config.json 提交到公开仓库
- 定期检查文件权限(`chmod 600 config.json`)

---

Dynamic import via __import__()

Medium
Category
Dangerous Code Execution
Confidence
75% confidence
Finding

Dynamic import() can load arbitrary modules at runtime, bypassing static analysis and potentially importing malicious code.

Content

Scanner excerpt · scripts/brush_step_skill.py (reported line 16)May include surrounding context.

python
def check_deps():
    try:
        __import__('requests')
        __import__('Crypto')
        return True
    except ImportError:

Dynamic import via __import__()

Medium
Category
Dangerous Code Execution
Confidence
75% confidence
Finding

Dynamic import() can load arbitrary modules at runtime, bypassing static analysis and potentially importing malicious code.

Content

Scanner excerpt · scripts/brush_step_skill.py (reported line 17)May include surrounding context.

python
def check_deps():
    try:
        __import__('requests')
        __import__('Crypto')
        return True
    except ImportError:
        print("[X] Missing deps. Run: pip install requests pycryptodome")

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The code deliberately fabricates IP addresses and is designed to support spoofed network origin during a login flow. In the context of a step-faking skill, this is not incidental metadata handling; it appears intended to evade anti-abuse, rate limiting, geolocation, or fraud-detection controls on the remote service.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill injects spoofed X-Forwarded-For and X-Real-IP headers into authentication requests using a fabricated address. Because this occurs during credential submission, it can help disguise abusive logins and defeat service-side trust assumptions around source attribution.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The request headers and payload force Chinese locale values such as "accept-language": "zh-CN" and "lang": "zh_CN". This is a natural-language policy issue because the code imposes a specific language/locale on users without any opt-in, alternative selection, or documented region-specific justification.

Content

No source excerpt is available for this finding.

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · scripts/huami.py (reported line 163)May include surrounding context.

python
# 4. Submit steps
    step = str(random.randint(step_min, step_max))
    today = datetime.datetime.now().strftime("%F")
    data_json = '%5B%7B%22data_hr%22%3A%22%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9L%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FVv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0v%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9e%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0n%5C%2Fa%5C%2F%5C%2F%5C%2FS%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0b%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F1FK%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FR%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9PTFFpaf9L%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FR%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0j%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9K%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FOv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fzf%5C%2F%5C%2F%5C%2F86%5C%2Fzr%5C%2FOv88%5C%2Fzf%5C%2FPf%5C%2F%5C%2F%5C%2F0v%5C%2FS%5C%2F8%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FSf%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fz3%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0r%5C%2FOv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FS%5C%2F9L%5C%2Fzb%5C%2FSf9K%5C%2F0v%5C%2FRf9H%5C%2Fzj%5C%2FSf9K%5C%2F0%5C%2F%5C%2FN%5C%2F%5C%2F%5C%2F%5C%2F0D%5C%2FSf83%5C%2Fzr%5C%2FPf9M%5C%2F0v%5C%2FOv9e%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FS%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fzv%5C%2F%5C%2Fz7%5C%2FO%5C%2F83%5C%2Fzv%5C%2FN%5C%2F83%5C%2Fzr%5C%2FN%5C%2F86%5C%2Fz%5C%2F%5C%2FNv83%5C%2Fzn%5C%2FXv84%5C%2Fzr%5C%2FPP84%5C%2Fzj%5C%2FN%5C%2F9e%5C%2Fzr%5C%2FN%5C%2F89%5C%2F03%5C%2FP%5C%2F89%5C%2Fz3%5C%2FQ%5C%2F9N%5C%2F0
...[truncated 27 chars]

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · scripts/huami.py (reported line 163)May include surrounding context.

python
# 4. Submit steps
    step = str(random.randint(step_min, step_max))
    today = datetime.datetime.now().strftime("%F")
    data_json = '%5B%7B%22data_hr%22%3A%22%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9L%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FVv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0v%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9e%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0n%5C%2Fa%5C%2F%5C%2F%5C%2FS%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0b%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F1FK%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FR%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9PTFFpaf9L%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FR%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0j%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9K%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FOv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fzf%5C%2F%5C%2F%5C%2F86%5C%2Fzr%5C%2FOv88%5C%2Fzf%5C%2FPf%5C%2F%5C%2F%5C%2F0v%5C%2FS%5C%2F8%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FSf%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fz3%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0r%5C%2FOv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FS%5C%2F9L%5C%2Fzb%5C%2FSf9K%5C%2F0v%5C%2FRf9H%5C%2Fzj%5C%2FSf9K%5C%2F0%5C%2F%5C%2FN%5C%2F%5C%2F%5C%2F%5C%2F0D%5C%2FSf83%5C%2Fzr%5C%2FPf9M%5C%2F0v%5C%2FOv9e%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FS%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fzv%5C%2F%5C%2Fz7%5C%2FO%5C%2F83%5C%2Fzv%5C%2FN%5C%2F83%5C%2Fzr%5C%2FN%5C%2F86%5C%2Fz%5C%2F%5C%2FNv83%5C%2Fzn%5C%2FXv84%5C%2Fzr%5C%2FPP84%5C%2Fzj%5C%2FN%5C%2F9e%5C%2Fzr%5C%2FN%5C%2F89%5C%2F03%5C%2FP%5C%2F89%5C%2Fz3%5C%2FQ%5C%2F9N%5C%2F0
...[truncated 27 chars]

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · scripts/huami.py (reported line 163)May include surrounding context.

python
# 4. Submit steps
    step = str(random.randint(step_min, step_max))
    today = datetime.datetime.now().strftime("%F")
    data_json = '%5B%7B%22data_hr%22%3A%22%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9L%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FVv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0v%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9e%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0n%5C%2Fa%5C%2F%5C%2F%5C%2FS%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0b%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F1FK%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FR%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9PTFFpaf9L%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FR%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0j%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F9K%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FOv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fzf%5C%2F%5C%2F%5C%2F86%5C%2Fzr%5C%2FOv88%5C%2Fzf%5C%2FPf%5C%2F%5C%2F%5C%2F0v%5C%2FS%5C%2F8%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FSf%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fz3%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F0r%5C%2FOv%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FS%5C%2F9L%5C%2Fzb%5C%2FSf9K%5C%2F0v%5C%2FRf9H%5C%2Fzj%5C%2FSf9K%5C%2F0%5C%2F%5C%2FN%5C%2F%5C%2F%5C%2F%5C%2F0D%5C%2FSf83%5C%2Fzr%5C%2FPf9M%5C%2F0v%5C%2FOv9e%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2FS%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2F%5C%2Fzv%5C%2F%5C%2Fz7%5C%2FO%5C%2F83%5C%2Fzv%5C%2FN%5C%2F83%5C%2Fzr%5C%2FN%5C%2F86%5C%2Fz%5C%2F%5C%2FNv83%5C%2Fzn%5C%2FXv84%5C%2Fzr%5C%2FPP84%5C%2Fzj%5C%2FN%5C%2F9e%5C%2Fzr%5C%2FN%5C%2F89%5C%2F03%5C%2FP%5C%2F89%5C%2Fz3%5C%2FQ%5C%2F9N%5C%2F0
...[truncated 27 chars]

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The manager passes raw usernames and passwords directly to an external routine that likely performs network authentication and step-modification actions, but the code provides no user-facing disclosure, consent checkpoint, or verification of how credentials are handled. In the context of a skill explicitly designed to alter fitness data across multiple accounts, this is more dangerous than a normal login flow because it encourages bulk submission of sensitive credentials to automation code that may use fake IPs and interact with third-party services in ways users may not fully understand.

Content

No source excerpt is available for this finding.