IdleClaw

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:14 PM UTC
• Version:
  1.2.1

Overview

IdleClaw does what it claims: it connects Ollama to a community inference network, with meaningful privacy and resource-use risks that are mostly disclosed and user-started.

Install only if you are comfortable with community inference. Do not send secrets, credentials, proprietary code, regulated data, or personal information through consume mode. In contribute mode, use a trusted or self-hosted IDLECLAW_SERVER if possible, monitor CPU/GPU and memory use, and stop the script when you no longer want to share local Ollama capacity.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 91% confidence
• Finding: The skill uses environment variables and network access (`IDLECLAW_SERVER`, `OLLAMA_HOST`, outbound HTTP/WebSocket communication) but does not declare corresponding permissions. This creates a transparency and trust problem: users may run a skill that can communicate off-host and consume local service data without an explicit permission prompt or review surface. In this context, the skill is specifically designed to expose local Ollama inference to a community network, which increases the sensitivity of undeclared network capabilities.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The script sends the user's prompt directly to a remote IdleClaw server for inference, but it does not clearly warn the user that prompt contents may leave the local machine and be processed by community-operated infrastructure. This creates a real confidentiality risk if users include secrets, proprietary code, or sensitive personal data in prompts, especially given the skill's stated purpose of using community inference.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: This node registers with a community server and accepts remote inference requests, then forwards the provided prompts/messages to the local Ollama instance without any explicit consent prompt or prominent warning that third-party prompt content will be processed on the user's machine. In a skill explicitly designed to 'share your idle Ollama inference with the community,' this behavior is expected, but it still creates a real privacy and trust risk because users may not understand that arbitrary remote prompts, including sensitive or abusive content, will transit through and be visible to their local system/operator.

VirusTotal

65/65 vendors flagged this skill as clean.