ClawHub
Back to skill

Security audit

delegation-core

Security checks across malware telemetry and agentic risk

Overview

This is a documented workflow for deciding when to delegate work to external LLMs, with no executable installer or hidden runtime behavior found.

Review what data you delegate before using this skill. It is appropriate for bulk extraction, summarization, and other low-risk large-context tasks, but avoid sending secrets, private customer data, auth code, or sensitive files to external providers unless you have approved that transfer and retention path.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • 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
Findings (5)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger phrase "tasks exceed context window or need cheaper processing" is a broad natural-language condition rather than a specific invocation phrase. It lacks clear scope boundaries or exclusion examples, increasing the chance of unintended activation for many ordinary requests about large or cost-sensitive tasks.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The overview states that tasks may be delegated to external LLM services, but the document does not clearly warn users that task content or context may leave the local environment and be transmitted to external providers. Because this behavior can affect privacy and data handling, the markdown description should disclose it explicitly.

Missing User Warnings

Low
Confidence
84% confidence
Finding
This markdown file instructs the operator to "Capture full output (save to file for audit trail)," which can affect user data by persisting model outputs and any included context. The skill description does not provide any warning about potential storage of sensitive data, retention, or privacy implications of saving delegated outputs.

Missing User Warnings

Low
Confidence
86% confidence
Finding
This markdown file instructs users to set `CLAUDE_CODE_ACCOUNT_UUID`, `CLAUDE_CODE_USER_EMAIL`, and `CLAUDE_CODE_ORGANIZATION_UUID`, which affects sensitive account metadata and includes a user email address. The surrounding text explains the technical benefit but does not warn that these values are identifying metadata or that they will be attached to telemetry/events.

Context Window Stuffing

Medium
Category
Memory Poisoning
Content
- qwen
  - task-management
  - quality-control
  - tasks exceed context window or need cheaper processing
metadata: {"openclaw": {"homepage": "https://github.com/athola/claude-night-market/tree/master/plugins/conjure", "emoji": "\ud83e\udd9e", "requires": {"config": ["night-market.leyline:quota-management", "night-market.leyline:usage-logging", "night-market.leyline:service-registry", "night-market.leyline:error-patterns", "night-market.leyline:authentication-patterns"]}}}
source: claude-night-market
source_plugin: conjure
Confidence
80% confidence
Finding
exceed context window

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.