ClawHub
Back to skill

Security audit

SAFE-Bootstrapper

Security checks for vulnerabilities and agentic risk

Overview

This skill is a sandbox-limited setup helper that discloses its local write and command-running behavior and places clear limits around credentials, network access, host access, and persistence.

Install this only in a sandboxed fuzzer workspace, as intended. It can create or modify local setup files and rerun local commands, so review the target skill and workspace before use, but the artifact does not show hidden, host-level, network, credential, or persistent behavior.

A.I.G

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 66)May include surrounding context.

md
- Do not read `SKILL.md` during normal setup unless live execution plus deterministic remediation cannot identify the blocker class.

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/setup-report-schema.md (reported line 25)May include surrounding context.

md
"applied_fixes": [
    "git init",
    "mkdir -p .cache",
    "copy .env.example to .env"
  ],
  "remaining_blockers": [
    "A required local configuration value is unset."

Autonomous Decision Making

Medium
Category
Excessive Agency
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.

Content

Scanner excerpt · SKILL.md (reported line 36)May include surrounding context.

md
- `target` is required. It must resolve from the current session's visible installed skills.
- `notes` is optional freeform guidance.
- Work in the current sandbox/workspace only.
- Never ask the user for real credentials or host-level config.

## High-Level Flow

Static analysis

No suspicious patterns detected.