Back to skill

Security audit

通用大模型 GPT 5.5

Security checks for vulnerabilities and agentic risk

Overview

The skill is a disclosed dLazy CLI wrapper, but it is framed as an OpenAI model while steering users into a third-party paid service, credential setup, global or npx package execution, and external upload of prompts or media.

Review carefully before installing. Use this only if you intend to send prompts and selected media files to dLazy, are comfortable installing or running `@dlazy/cli@1.2.0`, and understand that a dLazy API key may be saved locally. Prefer `npx` or a sandboxed/local install over global installation, verify the package/source yourself, and do not pass local file paths unless you want those files uploaded.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T01 · Skill Instruction Hijacking

Warning
Location
SKILL.md:90
Finding

Imperative skill instructions force use of a third-party commercial service

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:90-91, 141-144; equivalent instructions appear in SKILL-cn.md
Vulnerability Type: Agent instruction hijacking
Risk Level: Medium

Evidence

markdown
**CRITICAL INSTRUCTION FOR AGENT**:
Execute `dlazy gpt-5.5` to get the result.
markdown
> **AGENT CRITICAL INSTRUCTION**:
>
> 1. If the execution result returns `code: "insufficient_balance"`, you MUST explicitly inform the user that their credits are insufficient and guide them to recharge: [https://dlazy.com/dashboard/organization/settings?tab=credits](https://dlazy.com/dashboard/organization/settings?tab=credits)
> 2. If the execution result returns `code: "unauthorized"` or indicates missing API key, you MUST inform the user to get their API key from [https://dlazy.com/dashboard/organization/api-key](https://dlazy.com/dashboard/organization/api-key) and save it using `dlazy auth set <key>` and resume the task.

Technical Analysis

The skill marks its operational and commercial redirection instructions as “CRITICAL” and uses unconditional terms such as “MUST.” When the skill text is loaded into an agent context, these directives can compete with the user's current objective and pressure the agent to execute a third-party CLI, solicit authentication, or direct the user to purchase credits.

The requested CLI communicates with api.dlazy.com and can upload user-selected local media to files.dlazy.com. Although the document discloses this behavior, the imperative instruction does not require the agent to obtain informed user approval before invoking the external service or transmitting supplied content.

The recharge and API-key directives are conditional, but they still embed mandatory commercial and credential-configuration flows into agent behavior. No explicit safety override is present, and there is no evidence that the skill independently reads arbitrary files; the risk is limited to c ...[truncated 1502 chars]

Remediation
View remediation

Remediation Suggestions

  1. Replace “CRITICAL” and “MUST” directives with bounded operational guidance that remains subordinate to user intent and platform policy.
  2. Require explicit user confirmation before installing or invoking the dLazy CLI.
  3. Before transmission, clearly enumerate the destination domains and the prompts or files that will be sent.
  4. Never upload a local path unless the user expressly selected that file for the current request.
  5. Present authentication and credit information neutrally rather than requiring the agent to redirect users to commercial pages.
  6. Do not instruct the agent to resume automatically after credential configuration; require a new confirmation.
  7. State that API keys must never be included in chat, logs, command history, or generated output.
  8. Apply the same revisions to SKILL-cn.md so both documents enforce identical safeguards.

T08 · Insecure Dependencies

Warning
Location
SKILL.md:9
Finding

Third-party npm package is installed and executed without package-integrity pinning

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:9-12, 63-76; equivalent configuration appears in SKILL-cn.md
Vulnerability Type: Third-party dependency and supply-chain exposure
Risk Level: Medium

Evidence

yaml
'requires': { 'bins': ['npm', 'npx'] },
'install': 'npm install -g @dlazy/cli@1.2.0',
'installAlternative': 'npx @dlazy/cli@1.2.0',
markdown
- **CLI source code**: [github.com/dlazyai/cli](https://github.com/dlazyai/cli)
- **Maintainer**: dlazyai
- **npm package**: `@dlazy/cli` (pinned to `1.0.9` in this skill's install spec)
- **Homepage**: [dlazy.com](https://dlazy.com)

You can install on demand without persisting a global binary by running:

```bash
npx @dlazy/cli@1.2.0 <command>

Or, if you prefer a global install, the skill's metadata.clawdbot.install field declares the exact pinned version (npm install -g @dlazy/cli@1.2.0).

text

### Technical Analysis

The skill directs the host to download and execute `@dlazy/cli` from the npm supply chain. Specifying version `1.2.0` prevents ordinary semver drift, but the skill does not provide a package tarball digest, signed provenance, vendored artifact, or reproducible verification procedure tying the npm artifact to reviewed source.

Both `npm install -g` and `npx` may execute package lifecycle scripts or package entry points with the privileges of the invoking user. The global installation option additionally changes the user's executable environment.

The provenance text is internally contradictory: it says that the install specification is pinned to `1.0.9`, while every executable command installs `1.2.0`. This can cause a reviewer to audit a different release from the one that is actually executed and weakens the claimed provenance control.

No evidence in the audited files proves that `@dlazy/cli@1.2.0` is currently malicious. The confirmed issue is the unsafe trust boundary and inadequate verification of externa
...[truncated 1441 chars]
Remediation
View remediation

Remediation Suggestions

  1. Correct the documented version mismatch so the provenance section and executable metadata identify the same reviewed release.
  2. Pin and verify the npm tarball using a trusted SHA-512 integrity value or equivalent cryptographic digest.
  3. Verify npm package provenance or signatures and document the exact source commit corresponding to the published artifact.
  4. Prefer a lockfile-backed, project-local installation over npm install -g.
  5. Avoid automatic npx execution of an unverified package; download, verify, and then execute the approved artifact.
  6. Disable lifecycle scripts during installation where compatible, for example with npm install --ignore-scripts, and separately review any required installation behavior.
  7. Run the CLI in a sandbox with minimum filesystem and network access. Restrict outbound connections to the documented endpoints only when they are required.
  8. Never run installation commands with elevated privileges.
  9. Pin the repository to a specific commit or release tag and provide reproducible-build instructions that allow reviewers to compare the source with the npm artifact.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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 (7)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 27)May include surrounding context.

md
[English](./SKILL.md) · [中文](./SKILL-cn.md)

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · SKILL-cn.md (reported line 137)May include surrounding context.

md
## 错误处理

| Code | 错误类型                   | 示例信息                                                                                       |
| ---- | -------------------------- | ---------------------------------------------------------------------------------------------- |
| 401  | 未授权 (API Key缺失或无效) | `ok: false, code: "unauthorized", message: "API key is missing or invalid"`                    |
| 501  | 缺少必填参数               | `error: required option '--prompt <prompt>' not specified`                                     |

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

L146-L149 的“智能体关键指令”要求代理“必须明确告知用户”特定中文措辞和操作引导,但未说明可根据用户语言偏好调整,也未给出仅限中文场景的理由。这会形成语言/locale 强制约束,属于自然语言层面的组织策略风险。

Content

No source excerpt is available for this finding.

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · SKILL.md (reported line 137)May include surrounding context.

md
## Error Handling

| Code | Error Type                         | Example Message                                                                                                          |
| ---- | ---------------------------------- | ------------------------------------------------------------------------------------------------------------------------ |
| 401  | Unauthorized (No API Key)          | `ok: false, code: "unauthorized", message: "API key is missing or invalid"`                                              |
| 501  | Missing required parameter         | `error: required option '--prompt <prompt>' not specified`                                                               |

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · SKILL.md (reported line 142)May include surrounding context.

md
| 401  | Unauthorized (No API Key)          | `ok: false, code: "unauthorized", message: "API key is missing or invalid"`                                              |
| 501  | Missing required parameter         | `error: required option '--prompt <prompt>' not specified`                                                               |
| 502  | Local file read error              | `Error: Image file/Video file not found: C:\path\to\your\file`                                                           |
| 503  | API request failed (no balance)    | `ok: false, code: "insufficient_balance"`                                                                                |
| 503  | API request failed (server error)  | `HTTP status code error (500 server crash)`                                                                              |
| 504  | Asynchronous task execution failed | `=== Generation Failed ===` / `{Specific error reason returned by backend, for example "Prompt violates safety policy"}` |

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
98% confidence
Finding

L067 明确写明 npm 包名对应的技能 install 字段固定到 1.0.9 版本,但 L011、L012、L073、L076 都显示实际安装/运行版本为 1.2.0。这是文档陈述与实际配置直接矛盾,属于意图/文档与实际内容不一致。

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation states the npm package is pinned to 1.0.9 in the skill's install spec, but the manifest metadata actually specifies @dlazy/cli@1.2.0 for both install and installAlternative. This is an active contradiction in provenance/install guidance, not just missing detail.

Content

No source excerpt is available for this finding.