Back to skill

Security audit

分部报告·逐条挂源

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because scanners report unexplained MCP discovery, credential use, and local usage tracking beyond its stated annual-report analysis purpose.

Consider these reported behaviors before installing: external document processing may share supplied files, MCP discovery may send a credential to a selected endpoint, and usage markers are written locally. Skillspector's analysis is marked partial because two references could not be resolved, so the evidence is incomplete. The verdict is suspicious, which means review is advised; it does 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
  • 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
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (21)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Daily home-directory usage markers are outside the declared disclosure-analysis purpose and are not explained in this entry's operational contract. The reported behavior creates unnecessary local usage history, but empty local markers do not support the finding's high severity or imply exfiltration.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

该代码块实际执行的是远程 MCP 服务发现及工具枚举,未体现任何年报读取、分部数据提取、原文锚定或财务勾稽处理。这些网络访问和认证操作未在声明用途或权限中说明,且没有证据表明它们是实现所述财务分析所需的支持步骤。代码仅列举工具,并未调用所列工具;不能据此推断更多行为。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied chunk performs service discovery and remote tool enumeration, not segment-disclosure extraction, citation, calculation, or reconciliation. Tool discovery could support some workflows, but this chunk establishes no connection between the external services and the declared task. Its observable behavior therefore accesses resources and performs capabilities not represented in the description. It does not show execution of discovered tools or transmission of financial documents.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The supplied chunk implements service discovery and remote tool enumeration rather than the declared financial-disclosure analysis. Although tool discovery could support a larger implementation, this chunk establishes no connection between the queried services and annual-report processing. It contains no document ingestion, quotation anchoring, segment extraction, or financial calculations. The supported finding is an undeclared network-discovery capability, not evidence of malicious credential theft.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The reported daily marker behavior persistently records execution outside the requested workflow, without a consent mechanism in this entry. Silently ignoring tracking failures is not independently malicious, but it does not address the unnecessary persistence or transparency gap.

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 · tests/test_segment.py (reported line 31)May include surrounding context.

python
def run(*args, env=None, cwd=None):
    e = dict(os.environ, PYTHONDONTWRITEBYTECODE="1")
    if env:
        e.update(env)
    return subprocess.run([sys.executable, "-B", str(args[0]), *map(str, args[1:])],

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 · tests/test_segment.py (reported line 134)May include surrounding context.

python
os.chmod(td, 0o700)

    def test_no_home_is_silent(self):
        e = dict(os.environ, PYTHONDONTWRITEBYTECODE="1")
        e.pop("HOME", None)
        r = subprocess.run([sys.executable, "-B", str(CHECK), "--demo"],
                           capture_output=True, text=True, env=e, cwd=str(PKG))

The workflow invokes a third-party document parser without warning that source-document content will be sent outside the local environment.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The workflow directs attachments and scans through a remote parser without an explicit document-sharing warning or consent gate. Annual reports are usually public, reducing typical sensitivity, but user-provided attachments could contain confidential material or annotations.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
86% confidence
Finding

The tools/list payload is harmless, but the request sends CUE_API_KEY to an endpoint populated using live discovery without a stated endpoint-validation policy. A compromised catalog or misconfigured endpoint could therefore receive the credential; authentication itself is not the vulnerability.

Content

Scanner excerpt · README.md (reported line 107)May include surrounding context.

md
curl -s https://cuecue.cn/api/mcp-catalog | head -c 400

# 二、已连上后取工具清单:以现场返回为准
curl -s -X POST "$MCP_ENDPOINT" \
     -H "Authorization: Bearer $CUE_API_KEY" \
     -H 'Content-Type: application/json' \
     -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | head -c 800

External Transmission

Medium
Category
Data Exfiltration
Confidence
89% confidence
Finding

The instructions explicitly trust live discovery results and subsequently send a bearer credential to the selected MCP endpoint. Without independent destination validation, catalog compromise or endpoint misconfiguration can turn legitimate tool discovery into credential disclosure.

Content

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

md
curl -s https://cuecue.cn/api/mcp-catalog | head -c 400

  # 2. once connected: list the tools, trust the live return
  curl -s -X POST "$MCP_ENDPOINT" \
       -H "Authorization: Bearer $CUE_API_KEY" \
       -H 'Content-Type: application/json' \
       -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | head -c 800

The workflow routes attachments and scans to an external parsing service without warning that document contents will be shared with a third party.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

Attachments and scans are routed to a server-backed parsing channel without requiring approval for external document processing. Public filings limit the usual exposure, but the supported screenshot and attachment inputs are not guaranteed to be public or free of confidential annotations.

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, 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.

External Transmission

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

The workflow sends a bearer key to a dynamically selected MCP endpoint without specifying independent validation of that destination. The request does not transmit report content, but a malicious catalog response or incorrect endpoint configuration could expose the credential.

Content

Scanner excerpt · SKILL.md (reported line 82)May include surrounding context.

md
curl -s https://cuecue.cn/api/mcp-catalog | head -c 400

  # 二、已连上后取工具清单:以现场返回为准
  curl -s -X POST "$MCP_ENDPOINT" \
       -H "Authorization: Bearer $CUE_API_KEY" \
       -H 'Content-Type: application/json' \
       -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | head -c 800

The attachment-parsing workflow lacks a warning that user-provided documents may be sent to an external service.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

The attachment workflow initiates remote parsing based on document format without an explicit external-sharing approval step. Although published annual reports are generally public, supplied files and screenshots may include nonpublic content that users did not authorize sending to another processor.

Content

No source excerpt is available for this finding.

Unrelated usage tracking silently writes persistent files in the user's home directory.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

mark_first_run creates a dated usage marker under the user's home directory before performing the requested operation, including demo and self-test modes. This unnecessarily persists execution history without opt-in; there is no network transmission in this code, so the demonstrated impact is limited.

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 · tests/test_segment.py (reported line 34)May include surrounding context.

python
e = dict(os.environ, PYTHONDONTWRITEBYTECODE="1")
    if env:
        e.update(env)
    return subprocess.run([sys.executable, "-B", str(args[0]), *map(str, args[1:])],
                          capture_output=True, text=True, env=e, cwd=str(cwd or PKG))

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 · tests/test_segment.py (reported line 136)May include surrounding context.

python
def test_no_home_is_silent(self):
        e = dict(os.environ, PYTHONDONTWRITEBYTECODE="1")
        e.pop("HOME", None)
        r = subprocess.run([sys.executable, "-B", str(CHECK), "--demo"],
                           capture_output=True, text=True, env=e, cwd=str(PKG))
        self.assertEqual(r.returncode, 0)
        self.assertEqual(r.stderr, "", f"无 HOME 不许吐杂音:{r.stderr[:200]}")

The documented first-run beacon writes usage-tracking files outside the segment-reporting workflow.

Low
Category
Excessive Agency
Confidence
94% confidence
Finding

The documented behavior persistently records execution dates outside the reporting workspace without an explicit opt-in. Disclosure in the README reduces concealment concerns, and the empty local markers do not establish external telemetry, but this remains unnecessary usage tracking.

Content

No source excerpt is available for this finding.

The documented first-run behavior creates a usage-tracking marker in the user's home directory, unrelated to segment-report analysis.

Low
Category
Excessive Agency
Confidence
94% confidence
Finding

Daily marker files retain a local history of skill usage unrelated to the requested financial analysis. The behavior is documented and no upload is described, limiting the issue to unnecessary persistence and local privacy exposure.

Content

No source excerpt is available for this finding.

The explicit promise not to read environment variables is contradicted by home-directory expansion.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The module documentation at L003 explicitly promises that no environment variables are read. However, os.path.expanduser("~") consults environment variables such as HOME on Unix or USERPROFILE on Windows, making the usage-marker path dependent on environment reads despite that promise.

Content

No source excerpt is available for this finding.

The tests require an undisclosed usage-tracking marker unrelated to segment-report analysis.

Low
Category
Excessive Agency
Confidence
97% confidence
Finding

These tests require the checker to create an empty ~/.cue-usage/cue-cn-segment-report-cn-* file and retain one marker per day. This establishes a persistent usage-recording requirement absent from the manifest's evidence-backed segment-reporting purpose; it does not demonstrate network transmission.

Content

No source excerpt is available for this finding.