Back to skill

Security audit

品牌热点营销助手

Security checks for vulnerabilities and agentic risk

Overview

The skill does what it claims, but it handles a paid service credential in risky ways by saving it locally and sending it to one API over plain HTTP.

Review before installing. This skill may spend paid Zeelin credits, stores your App-Key locally for reuse, and includes an HTTP API example that could expose the key in transit. Use it only if you are comfortable managing that credential, prefer manual posting over automated Xiaohongshu publishing, and rotate/delete any key that may have been stored or sent insecurely.

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
SKILL.md:164
Finding

App-Key Transmitted Over Plaintext HTTP

Content
View full analysis
' \ --header 'Content-Type: application/json' ``` The descriptive label and placeholder above are English translations of the corresponding source text; the endpoint, header, and protocol are reproduced exactly. ### Technical Analysis The Skill directs the Agent to send the user's App-Key to an endpoint over unencrypted HTTP. An App-Key functions as a bearer credential: possession may be sufficient to authenticate requests and consume the associated account balance. HTTP provides neither transport confidentiality nor reliable endpoint authenticity. A network intermediary can inspect the `App-Key` header, modify the request or response, or redirect the Agent to attacker-controlled infrastructure. The use of `curl --location` further increases the importance of enforcing secure redirect behavior, because redirects must not be allowed to downgrade transport security or disclose credentials to an unexpected host. This behavior is unnecessary for the declared content-generation functionality. The same API operation should be provided exclusively over HTTPS. ### Attack Path 1. A user provides a valid Zeelin App-Key to the Skill. 2. The Skill sends a generation request to `http://vane.zeelin.cn/gateway/vane_marketing/content`. 3. An attacker positioned on the local network, an untrusted Wi-Fi network, a compromised router, or another intermediary observes or modifies the plaintext HTTP traffic. 4. The attacker extracts the value of the `App-Key` header. 5. The attacker replays the credential against the service to submit unauthorized requests or consume the victim's paid balance. ...[truncated 726 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:56
Finding

App-Key Persisted in an Insufficiently Protected Configuration File

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
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (8)

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill instructs the agent to read and reuse a previously stored App-Key from local persistent storage, which extends access beyond the immediate user request and creates cross-session secret retention. This increases the chance of credential exposure, accidental reuse for another user/task, or leakage from the local filesystem.

Content

No source excerpt is available for this finding.

Ssd 3

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The workflow explicitly instructs the agent to read and persist a user-provided App-Key across sessions, creating plain-language secret retention behavior. This is dangerous because it normalizes long-lived handling of a credential that can authorize billable actions and may be exposed through logs, files, or multi-user environments.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

Writing the validated App-Key into ~/.zeelin_config creates persistent credential storage unrelated to the narrow task of generating a marketing post. Persisting API credentials in a generic local file materially raises the risk of theft by other local processes, users, backups, or future unintended reuse.

Content

No source excerpt is available for this finding.

Ssd 3

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The instruction to write validated App-Keys to a local config file and overwrite previous keys creates a clear secret-retention mechanism in plain text workflow steps. Because the credential is then reused for future operations, compromise of that file could enable unauthorized spending and account misuse.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill tells the agent to persist a user's App-Key locally without a clear warning about credential-retention risks or filesystem exposure. Even if intended for convenience, storing secrets without transparent security guidance can lead to unsafe handling and user surprise.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
93% confidence
Finding

The skill transmits the user's App-Key to an external service and, in the content-generation example, uses plain HTTP rather than HTTPS for the base URL. Sending a sensitive credential over an unencrypted channel can expose it to interception, allowing unauthorized use of the user's account and quota.

Content

Scanner excerpt · SKILL.md (reported line 164)May include surrounding context.

请求示例

bash
curl --location --request POST 'http://vane.zeelin.cn/gateway/vane_marketing/content' \
--header 'App-Key: <用户的 zeelin_app_key>' \
--header 'Content-Type: application/json' \
--data-raw '{

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The extraction table says the topic-tag field comes from tags, but the documented API response examples use the singular field tag. This is an active contradiction within the skill documentation about what the code/process should extract, and it can lead the agent to omit or mishandle generated hashtags.

Content

No source excerpt is available for this finding.