Back to skill

Security audit

应收应付票据·逐条挂源

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because scanners report external attachment parsing without a clear privacy warning and hidden local usage tracking unrelated to its financial-document purpose; the evidence is incomplete because Skillspector recorded one unresolved reference.

The verdict is suspicious, so consider the reported attachment transfer, external service references, and automatic home-directory usage markers before installing. Skillspector scanned all listed components but marked its analysis partial because one local reference could not be resolved; the separate AIG scan completed without findings, which does not remove that evidence gap.

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
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
Findings (24)

YARA rule 'agent_skill_destructive_autonomous_actions': Autonomous destructive filesystem, shell history, or repository actions in AI agent skills [agent_skills]

High
Category
YARA Match
Confidence
75% confidence
Finding

YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).

Content

Scanner excerpt · README.en.md (reported line 137)May include surrounding context.

md
ly in a temporary directory and are deleted at exit.

The last line self-reports `档数=3|不合=N|不可判=N`; exit `0` all pass, `1` any mismatch, `3` value-domain failure.

`tests/` covers shell properties — `--help` exits 0, no argument exits 2, missing or empty file exits 3, `--demo` output passes its own gate and matches the good fixture byte for byte, and the first-run marker lands silently.

---

## 8. First-run marker

After argument parsing succeeds and before doing any work, the entry script drops an empty file at `~/.cue-usage/<slug>-<date>`.

**Why it exists:** both download counters on the platform are unreliable — a new listing is seeded with 50-60 initial counts in its first 48 hours, and that is not user downloads. This local file is the only evidence that somebody actually installed the package and ran it.

**It never gets in the way:** any failure (read-only home, quota full, no HOME) is swallowed silently — no error, no exit-code change, no noise. Runnin

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% 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 · README.en.md (reported line 149)May include surrounding context.

md
**It never gets in the way:** any failure (read-only home, quota full, no HOME) is swallowed silently — no error, no exit-code change, no noise. Running it several times a day leaves one file.

To clear it: `rm -rf ~/.cue-usage`.

---

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% 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 · README.md (reported line 149)May include surrounding context.

md
**它不会碍事:** 任何失败(家目录只读、配额满、无 HOME)都静默跳过,不报错、不改退码、不吐一句杂音。同一天跑多次只留一枚。

要清掉:`rm -rf ~/.cue-usage`。

---

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% 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 · README.en.md (reported line 149)May include surrounding context.

md
**It never gets in the way:** any failure (read-only home, quota full, no HOME) is swallowed silently — no error, no exit-code change, no noise. Running it several times a day leaves one file.

To clear it: `rm -rf ~/.cue-usage`.

---

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
85% 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 · README.en.md (reported line 149)May include surrounding context.

md
**It never gets in the way:** any failure (read-only home, quota full, no HOME) is swallowed silently — no error, no exit-code change, no noise. Running it several times a day leaves one file.

To clear it: `rm -rf ~/.cue-usage`.

---

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
85% 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 · README.md (reported line 149)May include surrounding context.

md
**它不会碍事:** 任何失败(家目录只读、配额满、无 HOME)都静默跳过,不报错、不改退码、不吐一句杂音。同一天跑多次只留一枚。

要清掉:`rm -rf ~/.cue-usage`。

---

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The Markdown checks support the declared financial-register workflow and are not themselves a mismatch. However, main() unconditionally invokes mark_first_run() after argument parsing. That function creates a directory in the user's home and touches a dated marker file, silently suppressing failures. This usage-recording side effect is not disclosed by the declared purpose and is unnecessary for the stated workflow. No network transmission is present in the supplied code.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

声明的核心功能是对财报中的票据事项进行有证据支持的信息整理;所示代码实际执行的是外部服务发现与工具枚举,未展示其与财报处理的关联,因此不能视为已说明的辅助实现细节。代码中的注释提到从目录获取路由地址,但实际片段没有设置 ROUTING_URL,其请求目标依赖外部环境。没有证据表明代码上传财报或执行已枚举的工具。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied code performs network-based service discovery rather than reading financial reports or producing evidence-linked bill registers. Although tool discovery could support a broader implementation, this chunk provides no connection between the external services and the declared financial-document task. It also does not derive ROUTING_URL from the catalog despite the comment suggesting that step. There is no evidence here of financial-data extraction or transmission.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

入口测试、演示与夹具校验属于支持声明用途的实现细节,不构成不匹配。但首次运行使用记录与财报票据整理无关,且访问并写入用户主目录中的隐藏位置,声明中未披露该行为。所给代码通过测试明确要求此行为,但未包含实际创建记录文件的实现;未发现网络传输或凭据访问证据。

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_notes.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_notes.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))

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.en.md (reported line 53)May include surrounding context.

CATALOG="https://cuecue.cn/api/mcp-catalog" curl -s "$CATALOG" | head -c 2000

take this group's routing address from the catalog

curl -s -X POST "$ROUTING_URL"
-H 'Content-Type: application/json'
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

text

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 53)May include surrounding context.

CATALOG="https://cuecue.cn/api/mcp-catalog" curl -s "$CATALOG" | head -c 2000

从目录里取本组的路由地址,确认工具在场

curl -s -X POST "$ROUTING_URL"
-H 'Content-Type: application/json'
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

text

Attachment parsing is initiated without a clear warning that document contents will be sent to an external service.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

The workflow directs attachment parsing through an external service when supporting quotes reside in attachments, without clearly stating that document contents leave the local environment or requiring approval. Public filings reduce the confidentiality risk, but user-supplied screenshots and attachments may contain nonpublic financial information, making implicit upload unsafe.

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.

The workflow sends user-provided attachments to an external parsing service without a privacy disclosure.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The workflow directs attachment parsing through an external service without requiring informed approval before transferring user-supplied material or explaining its data handling. Public exchange filings reduce the risk, but screenshots and private attachments may contain confidential financial or personal information that users did not authorize sharing with that provider.

Content

No source excerpt is available for this finding.

The financial-note checker creates persistent usage-tracking files unrelated to its stated purpose.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The checker unconditionally calls a routine that creates dated skill-usage markers in the home directory, including before checking whether a usable input path was supplied. These persistent artifacts reveal execution history without opt-in and are unrelated to document validation; the code shows no network transmission or storage of document contents.

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_notes.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_notes.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 usage marker is unrelated to the skill’s financial-document analysis purpose.

Low
Category
Excessive Agency
Confidence
92% confidence
Finding

The documented entry-point behavior silently persists the skill identifier and execution date outside the workspace without an opt-in. This creates an unnecessary local usage-history artifact that other parties with access to the home directory could inspect; there is no evidence here of remote telemetry or financial-document disclosure.

Content

No source excerpt is available for this finding.

The documented entry-point behavior adds persistent usage tracking unrelated to financial-report extraction or validation.

Low
Category
Excessive Agency
Confidence
90% confidence
Finding

The documentation describes automatic, silent persistence of execution dates and the skill identifier in the user's home directory. Although disclosed in the README, this tracking is unnecessary for financial-report validation and lacks an explicit opt-in, creating a limited local privacy exposure rather than demonstrated remote exfiltration.

Content

No source excerpt is available for this finding.

The module promises not to read environment variables but resolves the home directory through environment-dependent expansion.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The module documentation at L003 explicitly claims “不读环境变量” (does not read environment variables). However, os.path.expanduser("~") at L130 consults home-directory environment variables such as HOME on Unix, contradicting that guarantee even though the access is indirect.

Content

No source excerpt is available for this finding.

The tests require a persistent usage beacon unrelated to financial-note extraction or validation.

Low
Category
Excessive Agency
Confidence
97% confidence
Finding

The tests explicitly require the checker to create a persistent skill-specific usage marker under HOME and deduplicate it by day, enforcing a side effect unrelated to document validation. The tests use temporary homes, but the contract they enforce can leave usage-history metadata in real user homes; empty files and no demonstrated network transmission limit the impact.

Content

No source excerpt is available for this finding.