Back to skill

Security audit

Enterprise Bid Document AI — 企业招投标文书AI助手

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent China-focused bidding assistant that handles sensitive bid documents but discloses that risk and shows no hidden commands, exfiltration, or persistence.

Before installing, consider whether your organization permits third-party AI processing of bid materials. Redact confidential prices, client names, technical details, and personal data before use; have qualified staff verify outputs and current local rules; and prefer a reviewed or pinned ClawHub CLI instead of relying on an implicitly resolved npx package.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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 (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:720
Finding
Unpinned Third-Party Package Execution Through npx## Vulnerability Details **File Location**: `SKILL.md`, lines 720 and 733 **Vulnerability Type**: Insecure third-party dependency execution **Risk Level**: Medium ### Vulnerable Code Line 720: ```bash npx clawhub install tender-bidding-assistant@3.2.2 ``` Line 733: ```markdown **Install**: `npx clawhub install tender-bidding-assistant@3.2.2` ``` ### Technical Analysis The installation instructions pin the skill version to `3.2.2`, but they do not pin the version of the npm-distributed `clawhub` executable. When the command is invoked, `npx` may download and execute the package version currently resolved by the npm registry. The downloaded package and its potential lifecycle behavior are not included in this project, so their integrity and behavior cannot be established by this audit. Pinning the skill version does not protect the execution path from an altered, compromised, or unexpectedly updated `clawhub` package. This is a supply-chain weakness rather than evidence that the current artifact is itself malicious. ### Attack Path 1. An attacker compromises the npm publication account, package, or relevant dependency used by `clawhub`. 2. A user follows the documented installation command. 3. `npx` resolves and downloads the current unpinned `clawhub` package. 4. The downloaded package or its dependency lifecycle code executes with the invoking user's permissions. 5. Malicious code could access resources available to that user or modify the local environment. Exploitation depends on compromise or malicious modification of the external package supply chain. ### Impact Assessment Successful exploitation could provide code execution under the privileges of the user running the installation command. Depending on that user's permissions and environment, the affected scope could include: - Reading or modifying user-accessible files. - Accessing environment variables and locally available credentials. - Altering the working directory or user-level conf ...[truncated 228 chars]
Remediation
## Remediation Suggestions 1. Pin the `clawhub` executable to an exact, reviewed version: ```bash npx clawhub@<reviewed-exact-version> install tender-bidding-assistant@3.2.2 ``` 2. Record and verify the package integrity hash or lockfile metadata before execution. 3. Prefer a separately installed and reviewed CLI rather than allowing `npx` to perform an implicit download at invocation time. 4. Use a trusted internal registry or approved package mirror where appropriate. 5. Run installation in a restricted environment without sensitive credentials and with minimum filesystem and network permissions. 6. Document package provenance and a process for verifying that installed package contents match the reviewed source. 7. Consider disabling unnecessary npm lifecycle scripts during installation when compatible with the package.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Hidden Instructions

High
Category
Prompt Injection
Content
---
name: "Enterprise Bid Document AI — 企业招投标文书AI助手"
slug: tender-bidding-assistant
version: "3.2.2"
Confidence
60% confidence
Finding
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
- **Bid documents are confidential.** Uploaded files may contain pricing, technical proposals, and business-sensitive content.
- **Sanitize before upload.** Remove or redact sensitive data (specific prices, client names, core technical parameters, personnel identity information) before submitting documents for analysis.
- **No guaranteed deletion.** Uploaded documents are not guaranteed to be automatically deleted or anonymized after processing.
- **Your responsibility.** You are responsible for ensuring your organization's data security policies allow using third-party AI tools for business document analysis.

### Quick Start
Confidence
80% confidence
Finding
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
- **Bid documents are confidential.** Uploaded files may contain pricing, technical proposals, and business-sensitive content.
- **Sanitize before upload.** Remove or redact sensitive data (specific prices, client names, core technical parameters, personnel identity information) before submitting documents for analysis.
- **No guaranteed deletion.** Uploaded documents are not guaranteed to be automatically deleted or anonymized after processing.
- **Your responsibility.** You are responsible for ensuring your organization's data security policies allow using third-party AI tools for business document analysis.

### Quick Start
Confidence
80% confidence
Finding
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Natural-Language Policy Violations

Low
Confidence
82% confidence
Finding
The English README and security/compliance sections state that the skill covers China Government Procurement Law, China's Bidding Law, and related 2026 Chinese regulatory frameworks, but they do not explicitly say the skill is only for China-context procurements or require user confirmation when operating outside that locale. This can create a locale-policy issue because the skill's guidance is presented as generally applicable despite being tied to a specific jurisdiction.

Static analysis

No suspicious patterns detected.