Back to skill

Security audit

Geo Push Ops

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a Feishu notification helper, but it contains an undisclosed hardcoded webhook that can send messages to a fixed external destination when no configuration is provided.

Review this skill before installing. Use it only after removing or rotating the embedded Feishu webhook, requiring explicit webhook configuration, and validating that outbound destinations are limited to the intended Feishu endpoint. Treat any message passed to it as potentially sent outside your environment.

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 (2)

T09 · Insecure Skill Coding Practices

Error
Location
geo_push_ops.py:247
Finding

Hard-Coded Feishu Webhook Secret Causes Undisclosed Message Disclosure

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
geo_push_ops.py:266
Finding

Unvalidated Caller-Controlled Webhook Enables Server-Side Request Forgery

Content
View full analysis
DeliveryResult: """ 发送飞书消息(简化版 API) Args: message: 消息文本 webhook: Webhook URL max_retries: 最大重试次数 retry_delay: 重试延迟(秒) Returns: DeliveryResult 投递结果 """ config = FeishuConfig( webhook=webhook, retry_delay=retry_delay, max_retries=max_retries, ) return send_to_feishu(message, config) ``` The caller-controlled value is used directly as the request destination: ```python resp = requests.post(config.webhook, json=payload, headers=headers, timeout=10) ``` A validation function exists but is never enforced by either sending API: ```python def validate_webhook(webhook: str) -> bool: """ 验证飞书 webhook URL 格式 返回:是否有效 """ if not webhook: return False # 基本格式检查 if not webhook.startswith('https://open.feishu.cn/open-apis/bot/v2/hook/'): return False # 长度检查(应该有足够的随机字符) if len(webhook) < 80: return False return True ``` ### Technical Analysis `send_to_feishu_with_retry()` accepts its destination as a string, stores it in `FeishuConfig`, and forwards it to `requests.post()` without invoking `validate_webhook()`. Direct construction of `FeishuConfig` has the same problem because `send_to_feishu()` does not validate `config.webhook` either. If an integrating service allows an untrusted user to influence the webhook parameter, the module becomes a server-side request forgery primitive. The request originates from th ...[truncated 2363 chars]
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The natural-language description is written as an instruction/specification entirely in Chinese, with no indication that users may choose another language or that the skill is limited to a Chinese-only regulatory or regional context. Under the policy, forcing a specific language without user opt-in is a locale/language policy violation.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The module docstring describes the skill as handling Feishu push operations, implying a delivery-focused utility. However, substantial code implements domain-specific alerting behavior such as high-priority determination, market watch thresholds, bias text selection, and geopolitical/market message composition in lines L90-L230, which goes beyond pure push operations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The code embeds a live Feishu webhook URL and uses it as the default outbound destination, causing any caller of send_to_feishu() without explicit configuration to transmit message content to an external service. Hardcoded secrets/endpoints are dangerous because they can be abused by anyone with code access, enable unauthorized data exfiltration, and make accidental production notification leakage far more likely in a reusable skill context.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
95% confidence
Finding

The function performs outbound HTTP transmission of message content to a webhook using requests.post(). In this skill, external transmission is expected behavior, but it becomes security-relevant because message content may contain sensitive operational or market information and is sent to a remote endpoint with limited guardrails; combined with the default hardcoded webhook, the risk is elevated.

Content

Scanner excerpt · geo_push_ops.py (reported line 266)May include surrounding context.

python
start_time = time.time()
        
        try:
            resp = requests.post(config.webhook, json=payload, headers=headers, timeout=10)
            result.http_status = resp.status_code
            result.duration_ms = int((time.time() - start_time) * 1000)

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The file's module description and all user-facing output strings are written in Chinese, which effectively imposes a specific language on users. There is no indication that the skill is region-specific or that users can opt into a different language/locale.

Content

No source excerpt is available for this finding.