ClawHub
Back to plugin

Security audit

8080.ai

Security checks for vulnerabilities and agentic risk

Overview

The plugin is a coherent 8080.ai integration, but users should review it because it can start credit-consuming build actions, create share links for generated project artifacts, and has broader runtime routing than its skill guardrails suggest.

Install only if you intend OpenClaw to send project requirements, follow-up messages, selected media attachments, account/project queries, and build/agent actions to 8080.ai. Review the configured apiBaseUrl/siteUrl, be aware that build actions may use credits, and avoid sending sensitive project details unless you are comfortable with 8080.ai receiving and processing them.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

YARA rule 'agent_skill_prompt_injection_hidden_instructions': Prompt injection or hidden instructions embedded in AI agent skill text [agent_skills]

High
Category
YARA Match
Content
egisters the `ai8080_*` tools that this skill routes to.

## When this skill applies

Only route to `ai8080_*` tools when the user's message contains an explicit 8080.ai signal:
- the words "8080" or "8080.ai", OR
- an `/ai8080` command prefix, OR
- there is already an active 8080.ai project/session in this conversation.

Generic requests like "build me a todo app" or "help me build software with AI agents" do **not** qualify on their own — ask the user whether they want to use 8080.ai, or route to nothing, if the 8080.ai context is not already established.

## Tool routing guide

The phrases below only apply once the "When this skill applies" condition above is met.

| User says (with 8080.ai context established) | Tool to call |
|-----------|-------------|
| "login to 8080", "connect to 8080.ai" | `ai8080_login` with `action: "login"` |
| "set my 8080 api key to X", "my 8080 api-key is X", "use this 8080.ai api key X" | `ai8080_login` with `action: "set-api-key"`, `apiKey: X` |
| "
Confidence
80% confidence
Finding
YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
## Dashboard & Interactive UI

The 8080.ai plugin uses the OpenClaw Dashboard v2 features:
- **Project Selection**: Tools like `ai8080_list_projects` return an interactive dropdown. Do not ask the user for a project ID if they can select it from the UI.
- **Action Buttons**: After sending messages or starting projects, interactive buttons (Review, Continue, Run Agents) will appear in the chat. Tell the user they can click these buttons directly.
- **Automatic Context**: If a project is selected in the current session, you do not need to ask for a `projectId` for subsequent tool calls.
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.

Static analysis

No suspicious patterns detected.