Back to skill

Security audit

glimmer-delegation

Security checks for vulnerabilities and agentic risk

Overview

The skill is marked suspicious because its installation instructions pipe a downloaded installer directly into a shell, creating a significant supply-chain risk.

Review the installation step before installing. It runs a downloaded script without a separate inspection or independent integrity check, so consider using a trusted package manager or verifying the installer first, and understand and approve the system changes it may make. The scan was partial because two local path-like references could not be resolved, so the evidence does not cover everything those references might involve.

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
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (3)

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

The command downloads a mutable remote script and immediately executes it without inspection, version pinning, or independent integrity verification. Although the URL uses HTTPS and the expected Ollama domain, compromise of the distribution endpoint could execute arbitrary code, access local files, or undermine the skill's local-only privacy objective.

Content

Scanner excerpt · SKILL.md (reported line 48)May include surrounding context.

Installation

bash
curl -fsSL https://ollama.com/install.sh | sh
ollama pull muse-glimmer:30b

Chaining Abuse

High
Category
Tool Misuse
Confidence
96% confidence
Finding

The pipe into sh directly converts the remote response into executable shell instructions, removing an opportunity to review or verify the complete download before execution. This confirms the same unsafe installation mechanism reported by SC2, rather than a separate exploit or evidence of malicious command chaining.

Content

Scanner excerpt · SKILL.md (reported line 48)May include surrounding context.

Installation

bash
curl -fsSL https://ollama.com/install.sh | sh
ollama pull muse-glimmer:30b

Installation instructions execute a remote shell script without a system-integrity warning.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The installation instructions recommend executing a downloaded script without explaining that it can modify the system with the invoking user's permissions or any elevated permissions subsequently granted. Installing Ollama is relevant to this skill, but presenting both installation steps as required may encourage an agent to execute the installer without informed approval.

Content

No source excerpt is available for this finding.