Back to skill

Security audit

元宝派主动推送

Security checks for vulnerabilities and agentic risk

Overview

The skill appears to perform its stated Yuanbao messaging function, but it can send messages and upload selected local files to Tencent/Yuanbao using local bot credentials without built-in guardrails.

Install only if you trust the publisher and intend this agent to send Yuanbao group/private messages and upload chosen files to Tencent/Yuanbao infrastructure. Use a dedicated bot credential, restrict who can invoke the skill, add recipient and path allowlists where possible, avoid secrets or personal data in messages/uploads, and pin dependencies in a reviewed environment.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (2)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:40
Finding

Unpinned Third-Party Dependencies Create Supply-Chain Risk

Content
View full analysis
Remediation
View remediation
cos-python-sdk-v5== ``` 2. Generate and enforce cryptographic hashes for every package and transitive dependency: ```bash python3 -m pip install --require-hashes -r requirements.txt ``` 3. Commit the locked dependency file to the project and update `README.md` and `SKILL.md` to use it instead of direct unpinned installation commands. 4. Install dependencies in a dedicated virtual environment with only the permissions required by this Skill. 5. Review package provenance and release integrity before updating pinned versions. 6. Consider separating text-only messaging from file-upload support so that the COS SDK is installed only when file-upload functionality is required. ]]>

T09 · Insecure Skill Coding Practices

Warning
Location
send.py:350
Finding

Missing Server Acknowledgement Is Incorrectly Reported as Successful Delivery

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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 (9)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The README explicitly promotes proactive sending of messages and files to external Yuanbao/Tencent services, including direct WebSocket delivery and COS uploads, but it does not meaningfully warn users about the privacy and data-governance implications of transmitting message contents, file contents, recipient identifiers, and bot metadata to third-party infrastructure. In this skill’s context, that omission matters because the tool is designed for unattended cron/background use and cross-session notifications, which increases the chance that sensitive internal data is pushed externally without operator awareness or logging.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
97% confidence
Finding

The skill performs direct network operations over WebSocket and HTTPS/COS upload, but the manifest does not declare any tool scope such as permissions or allowed-tools. This creates a transparency and governance gap: an agent or reviewer may invoke a skill with powerful outbound messaging and file-transfer capability without an explicit permission boundary, increasing the risk of unauthorized data exfiltration, spam, or misuse of local credentials.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest description and the entire skill documentation are written only in Chinese, with no indication that users may choose another language or that the skill is intentionally limited to a Chinese-speaking audience for compliance or regional reasons. This creates a natural-language locale constraint that can violate organizational language-choice policies.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The script reads appKey/appSecret from the local OpenClaw config, exchanges them for a token over HTTPS, and then places the resulting token in WebSocket headers, but there is no user-visible notice in the CLI or runtime output that local credentials will be accessed and used for remote authentication. For a code-file warning check, the relevant sensitive operation lacks a prompt, visible disclosure, or nearby comment aimed at end users of the command.

Content

No source excerpt is available for this finding.

Cloud Storage Exfiltration

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

The code uploads arbitrary local files to Tencent COS using temporary cloud credentials, which is an external data transfer primitive. In a standalone utility this may be intentional, but in an agent skill it is security-relevant because any caller able to influence file_path can cause outbound transmission of local data to cloud storage.

Content

Scanner excerpt · send.py (reported line 508)May include surrounding context.

python
)
    client = CosS3Client(config)
    with open(file_path, "rb") as f:
        client.put_object(
            Bucket=upload_info["bucketName"],
            Body=f,
            Key=upload_info["location"],

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The file upload flows will transmit any local path supplied to the script to remote Tencent infrastructure without any confirmation, allowlist, or safety gate. In an agent/automation context, this increases the risk of unintended data exfiltration if another component passes sensitive local files such as SSH keys, config files, or documents.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

User-facing descriptions, help text, errors, and examples are written exclusively in Chinese, which effectively imposes a language choice on all users of the skill. The file does not offer any language selection, fallback, or opt-in mechanism that would make the locale restriction explicit and user-controlled.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The top-level documentation states that the skill directly sends group messages via the Yuanbao WebSocket protocol. However, the module also includes C2C/private messaging and file/image upload-and-send functionality, including non-WebSocket HTTP/COS operations. This is an active mismatch between the documented intent and the implemented scope.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The manifest description focuses on actively sending messages and files to Yuanbao group/private chats over WebSocket. In addition to messaging, the implementation performs separate HTTPS API calls to obtain sign tokens and upload credentials, then uploads local files to cloud object storage before sending file messages. While related to file delivery, this is broader behavior than the narrow 'direct WebSocket push' description suggests.

Content

No source excerpt is available for this finding.