Back to skill

Security audit

WorkBuddy 每日签到领积分

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because its reported key extraction, storage, and debugger activity exceed its stated daily check-in purpose and raise security concerns; some files were only partly inspected.

Given the reported secret-key handling, debugger behavior, and incomplete inspection, avoid installing unless you can verify those capabilities are necessary and acceptable. Treat the scan as a warning about real security concerns, not a finding that the author acted maliciously.

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 (1)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/attach_debug.js
Finding

Manual key extraction leaves the client's V8 inspector enabled

Content
View full analysis

Impact

A local attacker with loopback access can execute code with the client's OS privileges and access its session data. This is not an Internet-exposed listener by default.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
Findings (32)

Tainted flow: 'req' from os.environ.get (line 123, credential/environment) → urllib.request.build_opener.open (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/wb-checkin.py (reported line 129)May include surrounding context.

python
req.add_header("Content-Type", "application/json")
    req.add_header("Accept", "application/json")
    try:
        with _OPENER.open(req, timeout=20) as r:
            final_host = (urllib.parse.urlparse(r.geturl()).hostname or "").lower()
            if final_host != expected_host:
                return 0, {"raw": "refused: response from unexpected host %s" % final_host}

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

声明用途是 WorkBuddy 每日签到,而该脚本的直接目的和实际行为是提取、持久化并显示客户端存储加密密钥。这涉及未披露的敏感密钥访问及导出,不是普通签到请求的实现细节。代码没有展示密钥提取与签到的必要联系,也没有执行签到逻辑。所示脚本仅连接已开放的本地调试端口,并未自行开启端口;未发现向外部服务器传输密钥的行为。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied chunk implements key recovery rather than daily check-in requests or reporting. Obtaining and persisting a key for decrypting local credential storage is a security-sensitive capability not disclosed by the check-in description; it materially exceeds an ordinary supporting detail such as calling a documented check-in API. The helper implementations are not supplied, so their precise attachment, direct-extraction, and inspector-cleanup behavior cannot be verified. The visible code does not itself decrypt credentials or demonstrate external exfiltration.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared purpose is daily reward collection, but the supplied script's primary function is extracting and exporting key material associated with local credential encryption. This is a security-sensitive capability not disclosed in the description, and the chunk does not establish that it is necessary for check-in. No network exfiltration or actual credential decryption is shown; the mismatch is the undisclosed key extraction and storage, not those additional behaviors.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared purpose is daily check-in automation, but this chunk implements sensitive process inspection and secret extraction, neither of which is disclosed in the description. Although it targets WorkBuddy and could support a larger authentication workflow, extracting atRestSecretKey and returning the entire containing object is a material capability beyond the stated check-in behavior. The supplied code does not itself perform a check-in. It outputs the extracted data to stdout; no external exfiltration endpoint is shown.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The code performs process-debugging activation rather than a check-in operation. Enabling a debugger is a materially distinct capability not disclosed by the declared purpose, especially since the description says the interface and scripts are already established and client reverse engineering is unnecessary. The snippet does not demonstrate data extraction or further debugger activity, but the activation itself exceeds the declared check-in behavior.

Content

No source excerpt is available for this finding.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 33)May include surrounding context.

md
| 社区 1.2.0 | `buildkey_direct.js`:以 `ELECTRON_RUN_AS_NODE` 直接读客户端原生绑定取 build key,免调试端口、免找 PID、**客户端不必运行** |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

md
退出码:`0` = 成功领取 / 今日已领 / 活动未开启 / 仅查询;`1` = 失败。仅需 Python 3.8+ 标准库;`buildkey_direct.js` / CDP 兜底链路才需要 Node.js。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 78)May include surrounding context.

md
| `buildkey_direct.js` | **首选**免 GUI 取 build key(客户端不必运行) |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 181)May include surrounding context.

md
> `buildkey_direct.js` 已实测:客户端未运行也能取到 key,且与原先 CDP 取到的值一致。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 211)May include surrounding context.

md
- 本机凭据模式不需要客户端**正在运行**:有凭据 + build key 缓存即可;`buildkey_direct.js` 还能在客户端没开时补提 key

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 80)May include surrounding context.

md
| `buildkey_cdp.js` / `attach_debug.js` / `extract_buildkey.js` | CDP 兜底与人工排障链路 |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 80)May include surrounding context.

md
| `buildkey_cdp.js` / `attach_debug.js` / `extract_buildkey.js` | CDP 兜底与人工排障链路 |

Code capability 'env' detected in scripts/at_rest.py but not covered by declared permissions.

High
Category
MCP Least Privilege
Confidence
75% confidence
Finding

The skill uses 'env' capability that is not covered by the declared 'allowed-tools' preapproval. Review the host runtime's actual enforcement policy; runtime permissions are unknown unless separately verified, and this declaration alone does not establish a permission ceiling.

Content

No source excerpt is available for this finding.

Code capability 'network' detected in scripts/buildkey_cdp.js but not covered by declared permissions.

High
Category
MCP Least Privilege
Confidence
75% confidence
Finding

The skill uses 'network' capability that is not covered by the declared 'allowed-tools' preapproval. Review the host runtime's actual enforcement policy; runtime permissions are unknown unless separately verified, and this declaration alone does not establish a permission ceiling.

Content

No source excerpt is available for this finding.

The skill extracts and exports a client credential-decryption key, exceeding its stated daily check-in purpose.

High
Category
Excessive Agency
Confidence
96% confidence
Finding

The script extracts a credential-decryption key, persists it with no explicit restrictive permissions, and prints the complete key to stdout, exposing it to logs and potentially other local users. Obtaining authentication material may support check-in, but exporting a reusable decryption key expands exposure beyond the immediate operation and can enable decryption of accessible credential envelopes.

Content

No source excerpt is available for this finding.

Env Variable Harvesting

High
Category
Data Exfiltration
Confidence
60% confidence
Finding

Code enumerates, copies, or searches environment variables for secrets. Bulk environment access can collect credentials unrelated to the skill's stated purpose.

Content

Scanner excerpt · scripts/wb-checkin-guard.py (reported line 276)May include surrounding context.

python
uid = configured_uid()
    child_env = None
    if uid:
        child_env = os.environ.copy()
        child_env["WB_UID"] = uid

    argv = [exe] + list(FIXED_ARGV) + [script]

Several activation phrases are generic and do not identify WorkBuddy.

Medium
Category
Trigger Abuse
Confidence
97% confidence
Finding

Triggers such as “签到” (check in), “每日礼包” (daily gift), “checkin,” and “今天领了吗” (claimed it today?) can refer to unrelated services or ordinary conversation. Although the description identifies WorkBuddy, the trigger list does not require that context or provide exclusions, creating a risk of unintended activation.

Content

No source excerpt is available for this finding.

The 'Read-only' comment contradicts activation of the target process's debugger.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

Calling process._debugProcess changes the target's runtime state and can activate a debugger interface, contrary to the read-only label. That misleading description can cause an operator or agent to approve a sensitive operation as harmless inspection, while the helper leaves debugger lifecycle management unspecified.

Content

No source excerpt is available for this finding.

The script enables process debugging, a capability outside the daily check-in workflow.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The helper enables debugging on a caller-supplied PID, or an arbitrary hard-coded default, without checking that the target is WorkBuddy. For a compatible process accessible to the caller, this can expose a powerful inspector interface permitting code execution and secret access; this script provides no cleanup.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/buildkey_bootstrap.py (reported line 96)May include surrounding context.

python
def _fallback_tasklist():
    """兜底:tasklist 拿不到 PPID,只能按 PID 升序猜(不保证命中主进程)。"""
    try:
        r = subprocess.run(
            ["tasklist", "/FI", "IMAGENAME eq %s" % CLIENT_EXE, "/FO", "CSV", "/NH"],
            capture_output=True, text=True, encoding="latin1", errors="replace", timeout=20,
        )

Tainted flow: 'CLIENT_EXE' from os.environ.get (line 22, credential/environment) → subprocess.run (code execution)

Medium
Category
Data Flow
Confidence
65% confidence
Finding

Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.

Content

Scanner excerpt · scripts/buildkey_bootstrap.py (reported line 96)May include surrounding context.

python
def _fallback_tasklist():
    """兜底:tasklist 拿不到 PPID,只能按 PID 升序猜(不保证命中主进程)。"""
    try:
        r = subprocess.run(
            ["tasklist", "/FI", "IMAGENAME eq %s" % CLIENT_EXE, "/FO", "CSV", "/NH"],
            capture_output=True, text=True, encoding="latin1", errors="replace", timeout=20,
        )

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/buildkey_bootstrap.py (reported line 141)May include surrounding context.

python
def _run_cdp(node_exe, cdp_js, pid, port):
    try:
        r = subprocess.run(
            [node_exe, cdp_js, str(pid), str(port)],
            capture_output=True, text=True, encoding="utf-8", errors="replace",
            timeout=PER_PID_TIMEOUT,

Tainted flow: 'node_exe' from os.environ.get (line 236, credential/environment) → subprocess.run (code execution)

Medium
Category
Data Flow
Confidence
65% confidence
Finding

Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.

Content

Scanner excerpt · scripts/buildkey_bootstrap.py (reported line 141)May include surrounding context.

python
def _run_cdp(node_exe, cdp_js, pid, port):
    try:
        r = subprocess.run(
            [node_exe, cdp_js, str(pid), str(port)],
            capture_output=True, text=True, encoding="utf-8", errors="replace",
            timeout=PER_PID_TIMEOUT,

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/buildkey_bootstrap.py (reported line 169)May include surrounding context.

python
客户端**不必运行**、不占 9229 调试端口、不需要找主进程 PID。
    """
    try:
        r = subprocess.run(
            [node_exe, direct_js, "--json", "--out", out_path],
            capture_output=True, text=True, encoding="utf-8", errors="replace",
            timeout=45,