Security audit

外观设计检索

Security checks for vulnerabilities and agentic risk

Overview

The patent-search workflow is coherent, but API-key handling and cache-path validation need review before use.

Prefer a version that validates patent numbers and prevents credential forwarding across redirects. Until then, use ordinary patent numbers and the default HTTPS endpoint, avoid sharing confidential design details, and use a limited-spend key supplied outside chat. Review any persistent key storage and set a task budget.

A.I.G

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

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection

Findings (40)

Tainted flow: 'req' from os.environ.get (line 138, credential/environment) → urllib.request.urlopen (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/patentmax_client.py (reported line 142)May include surrounding context.
python
req.add_header("Content-Type", "application/json") req.add_header("Accept", "application/json") try: with urllib.request.urlopen(req, timeout=30) as response: payload = json.loads(response.read().decode("utf-8")) except urllib.error.HTTPError as exc: message = _server_message(exc.read().decode("utf-8", "replace"))

Tainted flow: 'req' from os.environ.get (line 333, credential/environment) → urllib.request.urlopen (network output)

Critical

Category: Data Flow
Confidence: 96% confidence
Finding: Sending the API key to the intended patent service is legitimate, but this request attaches it without enforcing HTTPS or restricting redirects. The default urllib redirect handling can forward Authorization to another origin, while an HTTP PATENTMAX_BASE_URL exposes the key in transit; exploitation requires an unsafe configuration or control of a redirect response.
Content: Scanner excerpt · scripts/patentmax_client.py (reported line 344)May include surrounding context.
python
req.add_header(key, value) try: with urllib.request.urlopen(req, timeout=TIMEOUT) as response: payload = response.read() # 服务端每个响应都回本次实扣和扣完后的余额。记下来，供 _spent/_balance 输出。 # 拿服务端的真值，而不是在本地累加估价：跨进程累加迟早算岔，

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 99% confidence
Finding: Using an API key could support patent searches, but printing part of the credential is not a necessary search implementation detail or a declared output. The supplied chunk exposes credential material rather than searching patents or producing comparisons. No network transmission is shown.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 97% confidence
Finding: The supplied command performs a substantive technical patent search, not the declared appearance-design search. Its query targets battery and electrolyte technology and contains no design-patent restriction or Locarno classification. The chunk also does not demonstrate image downloads or comparison-list generation, although their absence from this isolated snippet is not independently treated as a mismatch. Authentication and use of a patent-search API are supporting implementation details, not separate concerns.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 96% confidence
Finding: The supplied command targets /api/claims for an invention publication ending in A. Claims retrieval concerns technical patent protection, materially different from the declared appearance-design search and visual comparison workflow. Authentication is an ordinary implementation detail, not itself a mismatch. This assessment is limited to the supplied code chunk.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 95% confidence
Finding: The commands shown implement an invention-patent research workflow centered on a hard-coded publication. The Chinese publication suffix A identifies an invention application, not a design patent. Although similar-patent retrieval could support comparison, this chunk does not demonstrate the declared design-search workflow, representative-image downloads, or Excel-compatible comparison output. The mismatch is based on the different patent type and research operations, not merely on omitted output steps.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 95% confidence
Finding: The hard-coded query targets a technical subject materially different from the declared appearance-design search workflow. The command contains no design-patent or Locarno-class restriction. Image downloading and comparison-list generation are not shown, but their absence from this partial chunk is not independently treated as a mismatch.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 95% confidence
Finding: The image-download and Excel-compatible CSV export functions align with the description, and arbitrary search expressions could support design and Locarno queries. However, the executable CLI exposes substantial additional patent-research and analytics capabilities not declared in the narrow design-comparison purpose. Searches default to worldwide scope and do not enforce a design-patent restriction. Authentication, trial-key handling, billing controls, and local caching are supporting implementation details and are not the basis for this finding.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 94% confidence
Finding: Patent searching and image downloads overlap with the declared purpose, but the shown commands expose a materially broader patent-research workflow, including substantive text and citation analysis. The chunk does not demonstrate Locarno filtering or Excel-compatible comparison output; their absence alone is not the basis for the mismatch. This assessment is limited to the supplied command examples, not the unseen client implementation.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 93% confidence
Finding: Retrieving patent details could support the declared workflow, but this command specifically targets an invention-patent publication, not a design patent. That makes the resource and demonstrated operation inconsistent with the declared design-search purpose. The assessment is limited to the supplied invocation; the underlying client's implementation is not shown.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 88% confidence
Finding: The supplied command targets battery thermal-management technology rather than product appearance or design comparison. Nothing in the visible invocation constrains the search to design patents. It also does not demonstrate image downloads or an Excel-compatible comparison list, although those omissions alone would not establish a mismatch in a partial code chunk. The finding is limited to the visible command; the client's internal behavior is not supplied.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 88% confidence
Finding: The supplied workflow explicitly starts from core technical features and expands the closest patent semantically. This targets technical-solution prior-art retrieval rather than the declared appearance-design comparison workflow. Neither command visibly restricts results to design patents or uses Locarno classification. Image downloads and comparison-list generation are also not shown, although their absence alone in this partial chunk would not establish a mismatch.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

High

Category: MCP Tool Poisoning
Confidence: 87% confidence
Finding: Candidate tracking and CSV export are consistent supporting features. However, the explicit keep/exclude criteria target coolant technology, not appearance or Locarno classification. The cited patent identifiers also end in A, indicating invention application publications rather than Chinese design patents. This demonstrates a materially different screening scope. The chunk does not show searching or image downloading, but their absence alone is not treated as a mismatch.
Content: No source excerpt is available for this finding.

Code capability 'env' detected in scripts/patentmax_client.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/patentmax_client.py 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 search workflow sends user-provided product names and aliases to PatentMax without a privacy warning.

Medium

Category: Output Handling
Confidence: 91% confidence
Finding: These steps embed the user's product name and aliases in queries submitted to the third-party PatentMax service. Although the skill identifies its data provider and discloses costs, it does not warn that potentially confidential details about an unreleased product will be transmitted externally.
Content: No source excerpt is available for this finding.

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/api-reference.md (reported line 16)May include surrounding context.
bash
curl -s -H "Authorization: Bearer $PATENTMAX_API_KEY" \ "https://api.ip930.com/api/patent?pn=CN109761224A"

服务端自己完成内部寻址，这些前置调用不进用户账单。

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/api-reference.md (reported line 48)May include surrounding context.
服务端直接返回上次的结果、不再扣费；同一个键换了参数会被拒（400）；上一次还在处理中返回 409。

bash
curl -s -H "Authorization: Bearer $PATENTMAX_API_KEY" -H "Idempotency-Key: task42-search-1" "https://api.ip930.com/api/search?q=%E6%B5%B8%E6%B2%A1%E5%BC%8F%E6%B6%B2%E5%86%B7&ds=cn&size=20"

网络超时、不确定上一次有没有成功时，用同一个键重发，不要换键。

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/api-reference.md (reported line 89)May include surrounding context.
bash
curl -s -H "Authorization: Bearer $PATENTMAX_API_KEY" \ "https://api.ip930.com/api/patent?pn=CN109761224A"

某一块上游取不到时，该字段为 null 并在 unavailable 里列出，其余照常返回；

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/api-reference.md (reported line 114)May include surrounding context.
bash
curl -s -H "Authorization: Bearer $PATENTMAX_API_KEY" \ "https://api.ip930.com/api/claims?pn=CN109761224A"

只要同时需要两块以上，就用 /api/patent——它一口价，拆开单买反而贵。

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/tools.md (reported line 668)May include surrounding context.
试用用完、过期，或想用其他功能（查新、交底书等）时，命令会返回带 "free_trial": true 的错误，message 里写着下一步怎么办（注册 PatentMax 再送 ¥50 体验额度），原样转告用户，不要自己换写法重试。

用户有自己的密钥时用环境变量传，避免写进命令历史；设了就一律用它，不会领试用。密钥在 api.ip930.com 控制台创建：

bash
export PATENTMAX_API_KEY="pm_live_你的密钥"

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/api-reference.md (reported line 73)May include surrounding context.
curl -s -H "Authorization: Bearer $PATENTMAX_API_KEY"
--data-urlencode 'q=(固态电池 OR 全固态电池) AND 电解质'
--data-urlencode 'ds=all' --data-urlencode 'size=20'
-G "https://api.ip930.com/api/search"

text
> 检索式里有中文和括号，一定用 `--data-urlencode -G`，别手工拼 URL。

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/faq.md (reported line 35)May include surrounding context.
md
- **密钥复制不全。** 前后有没有多余空格、有没有漏掉尾部字符 - **少了 `Bearer ` 前缀。** 用 curl 时 header 必须是 `Authorization: Bearer pm_live_xxx`，只写密钥不行 - **密钥被撤销了。** 到 [控制台](https://api.ip930.com/features/api-platform)看状态 > 还有个容易踩的：如果请求里带了 `Authorization` 头，URL 上的 `?apikey=` 就不会被读取。两种方式选一种，别混用——带了一个格式错误的 header，正确的 URL 参数也会被忽略。

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/faq.md (reported line 41)May include surrounding context.
md
## 402 余额不足到[控制台](https://api.ip930.com/features/api-platform)充值。新账户的体验额度分两笔：注册送 ¥10，第一次调用成功后再送 ¥40。如果第二笔一直没到，确认第一次调用确实返回了 200——失败的调用不算。

External Transmission

Medium

Category: Data Exfiltration
Confidence: 50% 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 · references/faq.md (reported line 119)May include surrounding context.
md
## 花了多少钱脚本每次返回都带 `_spent`（本次实扣）和 `_balance`（余额），**这两个数来自服务端，是真值**。准确的流水到[控制台](https://api.ip930.com/features/api-platform)查。 | 调用 | 单价 | | --- | --- |