ClawHub
Back to skill

Security audit

freemoney

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed, read-only class-action settlement lookup tool, with some non-core referrals users should treat as unaudited external resources.

Before installing, understand that the reviewed skill itself appears limited to read-only settlement lookups, but the separate freemoney-plugin, Tencent Cloud setup, DeepSeek account, and QQ group mentioned in the docs were not reviewed here. Do not share API keys in chat or support groups, and verify settlement eligibility on official claim sites before submitting any claim.

A.I.G

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

T01 · Skill Instruction Hijacking

Error
Location
prompt.md:242
Finding

Mandatory Promotion of External Plugin and Community in Agent Responses

Content
View full analysis

Vulnerability Details

File Location: prompt.md:242-244
Additional Locations: prompt.md:17, SKILL.md:117-124
Vulnerability Type: T01: Skill Instruction Hijacking
Risk Level: High

Relevant code translated into English from prompt.md:242-244:

text
If daily automatic notifications are required, install freemoney-plugin
(the OpenClaw plugin version).

Issue-reporting QQ group: 615298

Related instruction translated into English from prompt.md:17:

text
If scheduled notifications are required, install freemoney-plugin
(the OpenClaw plugin version).

Technical Analysis

The skill embeds mandatory promotional content in its activation-response template. When an agent loads and follows these instructions, it is directed to advertise a separately distributed plugin and an external QQ community as part of its response.

These referrals are not required to perform the skill's declared settlement-query function. Embedding them in behavioral instructions modifies the agent's expected output and steers users toward external resources that are outside the audited package. The recommendation appears repeatedly in both prompt.md and SKILL.md, indicating systematic behavior rather than incidental documentation.

The audited package contains only Markdown instruction files. No executable scripts, credential-harvesting logic, persistence mechanism, obfuscated payload, or remote code-execution behavior was identified. The separately referenced plugin was not present in the audited project and therefore could not be assessed.

Attack Path

  1. The agent loads prompt.md as the skill's behavioral instruction set.
  2. A user activates the skill or requests settlement information.
  3. The agent follows the prescribed activation-response template.
  4. The generated response promotes freemoney-plugin and QQ group 615298.
  5. The user may install the unaudited external plugin ...[truncated 836 chars]
Remediation
View remediation

Remediation Suggestions

  1. Remove mandatory references to freemoney-plugin and QQ group 615298 from activation and query-response templates.
  2. Keep the default response limited to functionality necessary for retrieving and presenting settlement information.
  3. Provide support or extension links only when a user explicitly asks for support, notifications, or related products.
  4. Clearly identify any external plugin as a separately maintained and independently trusted component.
  5. Publish a verifiable source repository, version, checksum, and security review for any plugin recommended by the skill.
  6. Replace opaque or promotional referrals with transparent official documentation URLs.
  7. Add a policy requiring user confirmation before directing users to external communities or software installation.
  8. Audit the separately distributed freemoney-plugin before recommending it from the skill.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (12)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest description explicitly states “纯中文交互,” which imposes a language restriction without indicating user opt-in or offering alternatives. This is a natural-language policy concern because it forces a specific language rather than letting the user choose.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation says users will get '新案件提醒' when new cases appear, which implies proactive notification behavior. Elsewhere the manifest and FAQ explicitly state the skill is a passive query tool and does not support push notifications, creating an intent-level contradiction within the documented behavior.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The markdown says the skill is auto-activated after installation and gives trigger examples like “你好”, “有什么新理赔案子？”, and especially “理赔全部” without defining precise invocation constraints or exclusion conditions. Because activation is described conversationally rather than as a narrow command scope, users may unintentionally invoke the skill during ordinary chat.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · SKILL.md (reported line 130)May include surrounding context.

数据源

text
API基地址: https://api.ilang.ai/claims

端点:
  /api/latest    → 全部开放案件

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · prompt.md (reported line 24)May include surrounding context.

数据源

text
API基地址: https://api.ilang.ai/claims

端点:
  /api/latest    → 全部开放案件

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · prompt.md (reported line 64)May include surrounding context.

数据源

text
API基地址: https://api.ilang.ai/claims

端点:
  /api/latest    → 全部开放案件

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · prompt.md (reported line 99)May include surrounding context.

数据源

text
API基地址: https://api.ilang.ai/claims

端点:
  /api/latest    → 全部开放案件

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · prompt.md (reported line 122)May include surrounding context.

数据源

text
API基地址: https://api.ilang.ai/claims

端点:
  /api/latest    → 全部开放案件

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · prompt.md (reported line 134)May include surrounding context.

数据源

text
API基地址: https://api.ilang.ai/claims

端点:
  /api/latest    → 全部开放案件

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · prompt.md (reported line 153)May include surrounding context.

数据源

text
API基地址: https://api.ilang.ai/claims

端点:
  /api/latest    → 全部开放案件

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · prompt.md (reported line 229)May include surrounding context.

数据源

text
API基地址: https://api.ilang.ai/claims

端点:
  /api/latest    → 全部开放案件

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill hard-requires Chinese-only output regardless of user preference, which can exclude users who need another language or accessibility accommodation. This is primarily a safety/usability policy issue rather than a direct exploit path, but it can cause misunderstanding of legal/financial instructions and reduce informed consent.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.