Security audit
Security checks for vulnerabilities and agentic risk
The patent-analysis workflow fits its purpose, but credential handling, cache writes, and spreadsheet exports need security fixes.
Prefer a corrected version with safe credential storage, restricted redirects, contained cache paths, and formula-safe exports. Send only approved, nonconfidential search terms. Avoid sharing keys in chat, use a spending-limited key, import CSV columns as text, and check billing before retrying uncertain requests.
scripts/patentmax_client.pySpreadsheet formula injection through external patent metadata
Attacker-supplied formulas can alter displayed results or, in spreadsheet applications supporting external-request functions, initiate requests to attacker-controlled URLs. This finding does not establish arbitrary operating-system command execution.
Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.
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"))
Sending the PatentMax key to its intended API is legitimate, but this client attaches Authorization without enforcing HTTPS or restricting redirects. urllib's default redirect handling can forward that header to a different host, so a compromised endpoint or an insecure configured endpoint could expose the key and potentially sensitive patent-search queries.
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，
The supplied command exposes part of an API key rather than performing patent analysis or generating a technology breakdown table. Using an API credential could support the declared purpose, but printing its contents is a distinct, undeclared disclosure. The snippet shows no external transmission.
Search and statistics can support the declared workflow, and credential handling, caching, and budget controls are supporting implementation details rather than independent evidence of mismatch. However, this chunk exposes a substantially broader patent-research toolkit with several undeclared analysis and retrieval capabilities. It also lacks the central declared operation: constructing a technology hierarchy and exporting branches with keywords, queries, and patent counts. The supplied code therefore only partially supports, rather than accurately represents, the declared purpose. This assessment is limited to the supplied chunk; other skill components could implement the missing workflow.
The commands operate on individual patent candidates and their review statuses, rather than technology branches, keywords, queries, or branch-level counts. Although candidate review is related to patent analysis, it is a distinct workflow not described by the declared technology-decomposition purpose. This assessment is limited to the supplied commands; the underlying client implementation is not shown.
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.
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.
The workflow directs the agent to submit the user's technology description and derived queries to an external service without first checking whether they contain confidential invention details. Naming PatentMax as the data source provides some transparency, but patent-analysis inputs can disclose unpublished inventions or research priorities, making an explicit confidentiality check important.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
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。
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
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；
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
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。
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
- **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。
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
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。
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。
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
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"
> 检索式里有中文和括号，一定用 `--data-urlencode -G`，别手工拼 URL。
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
## 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。
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
## 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 |
| --- | --- |
The workflow directs the agent to construct patent-search queries from the user's technical proposal without warning that those terms will be transmitted to an external service. In a patent-analysis context, queries can reveal unpublished inventions or commercially sensitive technical details, creating a confidentiality risk even though external searching is a legitimate part of the task. The provided file does not establish the client's transmission behavior or whether consent safeguards exist elsewhere, so the finding is limited to the missing disclosure and confidentiality check in this workflow.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
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：
export PATENTMAX_API_KEY="pm_live_你的密钥"
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -s "https://api.ip930.com/api/v1/health"
不需要密钥，用来确认网络通不通。
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# 设了 PATENTMAX_API_KEY 就一律用它，不会去领试用。
TRIAL_SOURCE = "patent-search-cn"
TRIAL_GROUP = "patent-search"
TRIAL_REGISTER_URL = "https://api.ip930.com/features/api-platform"
KEY_FROM_TRIAL = False
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
if _billing["balance_cents"] < 500 and not KEY_FROM_TRIAL: # 试用用户不提醒，用完时服务端会给注册指引
out["_balance_warning"] = (
"余额不足 ¥5，继续批量检索会很快耗尽。"
"请先提醒用户到 https://api.ip930.com/features/api-platform 充值。"
)
if _billing["free_quota"]:
left = max(0, _billing["free_quota"] - (_billing["free_used"] or 0))
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
hint = {
400: "参数有问题，检查必填项和取值范围。",
401: "密钥无效或已撤销。确认 PATENTMAX_API_KEY 是 pm_live_ 或 pm_test_ 开头的完整密钥。",
402: "余额不足。到 https://api.ip930.com/features/api-platform 充值。",
403: "该密钥没有此接口的权限。",
404: "没找到这篇专利。确认公开号正确且带国别代码，如 CN109761224A。",
409: "同一个请求上一次还在处理中。等几秒重跑同一条命令即可，不会重复扣费。",
An environment-configured cache directory is not inherently unsafe, but this operation writes an API credential before restricting its permissions with chmod. With a permissive umask, the key can briefly be readable by other local users, and ordinary open also follows preexisting symlinks; exploitation depends on local filesystem access and permissions.
path = _trial_key_path()
try:
os.makedirs(os.path.dirname(path), exist_ok=True)
with open(path, "w", encoding="utf-8") as fh:
json.dump({
"api_key": key,
"expires_at": data.get("expires_at"),
The network-error guidance broadly assures the agent that retrying will not charge twice, while the cache documentation explicitly states that whole-patent responses are not replay-protected server-side. If a request is billed but its response is lost, no local cache is created, and following this advice can incur another charge.
Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.
def _save_ledger(ledger):
try:
with open(_ledger_path(), "w", encoding="utf-8") as fh:
json.dump(ledger, fh, ensure_ascii=False)
except OSError:
pass