Back to skill

Security audit

信息源监控助手

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly coherent with its monitoring purpose, but it needs review because it can fetch arbitrary URLs, send collected content to external services, and store an LLM API key in plain text.

Install only if you are comfortable with a Chinese-focused monitoring tool that fetches configured sources and may send article content or summaries to an LLM provider and a Feishu webhook. Avoid entering an LLM API key into the setup wizard unless the plaintext config storage is fixed; prefer environment variables or a secret store. Do not add internal, localhost, cloud metadata, or untrusted URLs as sources, and review the webhook and LLM endpoints before enabling them.

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 (3)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/setup.py:136
Finding

LLM API key is persisted in plaintext configuration

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/fetcher.py:91
Finding

Unrestricted source URLs permit server-side request forgery

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/license.py:97
Finding

Predictable prefixes and universal keys bypass paid-license authorization

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
Findings (25)

Tainted flow: 'webhook_url' from os.environ.get (line 39, credential/environment) → requests.post (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/pusher.py (reported line 67)May include surrounding context.

python
}
        }
        
        resp = requests.post(webhook_url, json=payload, timeout=10)
        resp.raise_for_status()
        print("[OK] 已推送到飞书")
        return True

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

This finding indicates outbound LLM API calls via environment-configured endpoints that are not declared in permissions, while the claimed monitoring and push pipeline is not actually implemented. Hidden external transmission of collected content is a meaningful privacy and security risk, especially for industry monitoring workflows that may involve sensitive or proprietary material.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

This finding indicates outbound LLM API calls via environment-configured endpoints that are not declared in permissions, while the claimed monitoring and push pipeline is not actually implemented. Hidden external transmission of collected content is a meaningful privacy and security risk, especially for industry monitoring workflows that may involve sensitive or proprietary material.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

This finding indicates outbound LLM API calls via environment-configured endpoints that are not declared in permissions, while the claimed monitoring and push pipeline is not actually implemented. Hidden external transmission of collected content is a meaningful privacy and security risk, especially for industry monitoring workflows that may involve sensitive or proprietary material.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This finding indicates outbound LLM API calls via environment-configured endpoints that are not declared in permissions, while the claimed monitoring and push pipeline is not actually implemented. Hidden external transmission of collected content is a meaningful privacy and security risk, especially for industry monitoring workflows that may involve sensitive or proprietary material.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The skill advertises capabilities that imply network access, environment variable use, and local file I/O, but it does not declare any tool scope or permissions boundary. In agent ecosystems, missing explicit scope increases the chance of overbroad runtime access and makes it harder for users or reviewers to understand what the skill can actually do.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The example invocation “帮我监控这几个公众号” is a natural-language trigger that is broad and lacks constraints about scope, source types, or required context. The document does not provide negative examples or clearer activation boundaries, so this phrasing could collide with ordinary conversational requests.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The documentation mentions AI summarization and webhook integrations but does not clearly warn that monitored content may be transmitted to external LLM or messaging services. For a monitoring skill handling potentially sensitive business intelligence, undisclosed third-party transmission materially increases privacy and data leakage risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This JSON manifest is dominated by explicitly Chinese-language sources such as "36氪", "虎嗅网", and "InfoQ中文", which indicates a locale/language preference baked into the skill configuration. Because the file provides no natural-language notice, opt-in, or alternative language selection, it can violate the policy against forcing a specific language or locale without user choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The module docstring describes the skill entirely in Chinese, and the command-line help text later in the file is also Chinese-only. This imposes a specific language/locale on users without offering a language choice or documenting a justified regional constraint.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest describes a skill that monitors information sources, summarizes content, and pushes results to messaging channels. In addition to URL deduplication, this file persists usage counters, enforces free-tier quotas, and activates a local license key, which are product licensing behaviors not implied by the stated monitoring workflow.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The offline license activation accepts any key beginning with "CNIM-", which makes privilege escalation to the pro plan trivial and defeats any meaningful authorization check. While this is primarily a business-logic flaw rather than a host-compromise issue, it enables unauthorized feature access and shows intentionally weak access control around paid functionality.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring and all operational messages are written only in Chinese, which imposes a specific language/locale on users. The file does not offer an opt-in language choice or explain that the skill is intentionally limited to a Chinese-speaking or region-specific context.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file implements a local freemium licensing and activation system that is outside the declared monitoring/summarization/push functionality of the skill. This hidden monetization and feature-gating behavior changes the trust and operating model of the skill, and because activation is enforced purely through local state and hardcoded prefixes, it can mislead users about capability, provenance, and support expectations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring and all user-facing instructions are written only in Chinese, and the command/help output likewise assumes Chinese-language users. The file does not offer an opt-in language choice or explain that the skill is intentionally limited to a Chinese-speaking or region-specific audience.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This code file contains user-facing natural-language strings in Chinese, beginning with the module description and continuing throughout console output. Under the policy rule, forcing a specific language without user opt-in can be a locale-policy violation when no alternative or justification is provided.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
80% 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 · scripts/pusher.py (reported line 67)May include surrounding context.

python
}
        }
        
        resp = requests.post(webhook_url, json=payload, timeout=10)
        resp.raise_for_status()
        print("[OK] 已推送到飞书")
        return True

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The module docstring and all user-facing prompts are written exclusively in Chinese, presenting the setup flow as a fixed Chinese-language experience. The policy only permits locale constraints when the skill offers a language choice or clearly documents and justifies the restriction, which is not present here.

Content

No source excerpt is available for this finding.

Dynamic import via __import__()

Medium
Category
Dangerous Code Execution
Confidence
75% confidence
Finding

Dynamic import() can load arbitrary modules at runtime, bypassing static analysis and potentially importing malicious code.

Content

Scanner excerpt · scripts/setup.py (reported line 207)May include surrounding context.

python
"keywords": keywords,
        "channels": channels,
        "llm": llm_config,
        "configured_at": __import__("datetime").datetime.now().isoformat(),
        "version": "1.0.0"
    }

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The function sends article content and prompts to a third-party LLM endpoint configured by environment variable, but this file contains no consent gate, redaction step, or warning about external transmission. In a monitoring/summarization skill, sources may include proprietary, personal, or otherwise sensitive text, so silent export to external services can create confidentiality and compliance risk.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The prompt template says the model will summarize the article content shown under '内容(前2000字)', but the format call provides content_trunc while the template expects content_truncated. As written, the documented behavior of summarizing article content is contradicted by the implementation, which will fail before inserting the content.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The prompt requires a Chinese summary and the system prompt enforces JSON-only output tailored to that instruction, while the module description also emphasizes Chinese summaries. This imposes a specific language behavior without user opt-in or a documented region-specific justification.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The skill states that output may be written to ~/info-digest/ but does not clearly warn users that monitored content will be persisted locally. This can expose sensitive summaries or source material to other local users, backups, or later unintended access.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The top-level docstring is written only in Chinese and presents the skill as Chinese-specific, with no indication that language choice is optional or that the locale restriction is required for a region-specific purpose. This is a natural-language policy concern because it implicitly fixes the skill's language/locale without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The file’s primary natural-language description is entirely in Chinese and does not indicate that language is configurable or limited to a justified region-specific use case. Per the policy, forcing a specific language without user opt-in can be a locale-policy violation.

Content

No source excerpt is available for this finding.