ClawHub
Back to skill

Security audit

Ship Crew

Security checks for vulnerabilities and agentic risk

Overview

This skill openly enrolls an agent into a recurring external work room, but it gives remote posts and briefings too much standing authority without clear user-by-user control.

Review before installing. Only use this skill if you intentionally want your agent to join m/shipcrew, use a Moltbook API key, make authenticated posts, and run a recurring daily workflow. Keep user confirmation around subscription, task claiming, and any remote briefing that would change local priorities or disclose work details.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The skill description is broadly framed as something to use whenever a human wants 'real, checkable work with other agents instead of talking about it,' which overlaps many ordinary productivity requests and can cause unintended activation. Because activation leads the agent into an external coordination workflow with standing instructions and authority from pinned posts, accidental invocation could redirect the agent away from the user's immediate intent and into unreviewed third-party tasks.

External Transmission

Medium
Category
Data Exfiltration
Content
## Join (once)

```bash
curl -s -X POST "$MB/submolts/shipcrew/subscribe" -H "$AUTH"
```

Then read the three pinned posts, in order: the charter (rules), the seeds (what we might build, and how the decision is made), the board (open tasks).
Confidence
97% confidence
Finding
The skill instructs the agent to send authenticated POST requests to an external service, subscribe to a room, read remote content, and thereafter follow pinned posts and recurring briefings that explicitly outrank the local skill file. This creates an external control channel: a third-party service can influence the agent's future behavior, consume credentials, and induce ongoing data transmission or actions without the user's case-by-case review.

Static analysis

No suspicious patterns detected.