Feapder 中文助手

• Outcome: Review
• Latest audit: May 29, 2026 · 8:21 PM UTC
• Version: 1.0.0

Overview

The skill is a coherent feapder coding helper, but users should be careful because its bundled upstream reference files include unsafe Redis parsing and credential-handling examples.

Install only if you want a feapder 1.9.2 coding reference. Do not run vendored reference code blindly, do not paste real cookies or passwords into examples, avoid untrusted Redis task/cache data, and keep debug Redis keys and database writes isolated from production.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (10)

eval() call detected

High

• Category: Dangerous Code Execution

Content

def get_task_from_redis(self):
        tasks = self._redisdb.zget(self._task_table, count=self._task_limit)
        tasks = [eval(task) for task in tasks]
        return tasks

    def get_todo_task_from_mysql(self):

• Confidence: 99% confidence
• Finding: The code deserializes Redis task entries with Python eval(), which executes arbitrary Python expressions rather than safely parsing data. If an attacker can influence the Redis task payload or seed table contents, they can achieve arbitrary code execution in the spider process, making this a genuine and severe vulnerability.

eval() call detected

High

• Category: Dangerous Code Execution

Content

log.info("无response缓存  重新下载")
            response_obj = self.get_response(save_cached=save_cached)
        else:
            response_dict = eval(response_dict)
            response_obj = Response.from_dict(response_dict)
        return response_obj

• Confidence: 99% confidence
• Finding: The code calls eval() on data loaded from Redis before converting it into a Response object. If an attacker can influence the Redis value, this enables arbitrary Python code execution in the crawler process, which is far more dangerous than simple deserialization of cached response data.

Intent-Code Divergence

Medium

• Confidence: 94% confidence
• Finding: DebugSpider emits a warning stating that debug mode will not write data, but when save_to_db=True it only skips replacing ITEM_PIPELINES and therefore can retain the application's normal persistence pipelines. This creates a misleading safety guarantee: developers may run a supposedly non-persistent debug session against production-like settings and unintentionally write records or trigger side effects in downstream storage.

Context-Inappropriate Capability

High

• Confidence: 99% confidence
• Finding: Response cache loading uses Redis-backed data as executable input via eval(), creating a direct code-execution path from cache storage into application runtime. In a scraping framework, Redis is often shared infrastructure, so compromise or poisoning of cache entries can lead to full process takeover.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The documentation explicitly shows supplying a database password on the command line. Command-line arguments are commonly exposed through shell history, process listings, audit logs, and screenshots, so this can leak credentials to other local users or logging systems. In the context of a developer CLI guide, this is more dangerous because readers are likely to copy-paste the example directly into real environments.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The documentation recommends storing MySQL credentials in environment variables but does not mention that these values are sensitive or how to protect them. Environment variables are often inherited by child processes, can be exposed in CI/CD logs, crash dumps, shell init files, and shared user environments, so presenting this without safeguards promotes insecure secret handling. The skill context increases risk because it is operational guidance aimed at developers configuring real database access.

Missing User Warnings

Medium

• Confidence: 98% confidence
• Finding: The curl example contains a full Cookie header with what appears to be live session tokens and gives no warning to redact authentication material before sharing or pasting requests. Session cookies can enable account/session hijacking if reused, and documentation that normalizes copying such data into terminals or tickets materially increases accidental credential exposure. In a debugging skill, this is especially risky because users are encouraged to paste captured traffic from browsers and proxies.

Missing User Warnings

Medium

• Confidence: 80% confidence
• Finding: Debug mode forces DELETE_KEYS=True in __debug_custom_setting__, and later run() deletes tables matching self._redis_key + "*". Because __init__ derives the debug redis key by appending "_debug" to a caller-provided redis_key without validating scope, this can cause broad Redis key cleanup that may remove more state than an operator expects, especially in shared Redis environments.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: This finding is the same underlying issue as the eval() call: Redis task data is treated as executable Python code without disclosure or safety controls. In a spider framework, task payloads often originate from external queues or shared systems, so unsafe deserialization in this context materially increases the likelihood of remote code execution.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The debug log prints the full request URL, method, and request arguments, which can include headers, cookies, auth data, and proxy settings. This can expose session tokens, credentials, or other sensitive data into logs, creating a secondary disclosure channel even if the network request itself is legitimate.

VirusTotal

66/66 vendors flagged this skill as clean.

Static analysis

Detected: suspicious.dynamic_code_execution

Dynamic code execution detected.

Critical

• Code: suspicious.dynamic_code_execution
• Location: references/vendor/feapder-1.9.2/feapder/core/spiders/task_spider.py:360

Dynamic code execution detected.

Critical

• Code: suspicious.dynamic_code_execution
• Location: references/vendor/feapder-1.9.2/feapder/network/request.py:527