Back to skill

Security audit

glm-delegation

Security checks for vulnerabilities and agentic risk

Overview

The skill is marked suspicious because scanners flag submitting local documents to an external model without a privacy warning and selecting a model provider, while review coverage is partial.

Before installing, check whether you are comfortable with local documents being sent to an external model provider and with the selected provider potentially using a different account or billing plan. The executor implementation was not included, and Skillspector reported partial coverage because several references could not be resolved, so the evidence is incomplete.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

External Model or Provider Selection

High
Category
Excessive Agency
Confidence
90% confidence
Finding

Skill selects an external model or provider that may use a different account or billing plan than the operator expects. Undisclosed model switches can cause unexpected cost or quota consumption.

Content

Scanner excerpt · SKILL.md (reported line 88)May include surrounding context.

bash
ANTHROPIC_BASE_URL=https://api.z.ai/api/anthropic \
ANTHROPIC_AUTH_TOKEN="$ZAI_API_KEY" \
claude --model glm-5.3 -p "Review this design"

Smart Delegation

The quick-start command supplies local documents to an external model service without a privacy warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The quick-start example supplies a local directory to an external model without instructing the agent to review its contents or establish that disclosure is authorized. Although external delegation is the skill's stated purpose, directory-wide submission can inadvertently disclose confidential documents, especially when provider selection is automated; the executor implementation is not provided to establish whether it applies safeguards.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 86)May include surrounding context.

Direct CLI usage

bash
ANTHROPIC_BASE_URL=https://api.z.ai/api/anthropic \
ANTHROPIC_AUTH_TOKEN="$ZAI_API_KEY" \
claude --model glm-5.3 -p "Review this design"

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 103)May include surrounding context.

md
| Property | Value |
|----------|-------|
| Binary | `claude` (stock) |
| Base URL | `https://api.z.ai/api/anthropic` |
| Auth variable | `ANTHROPIC_AUTH_TOKEN`, from `ZAI_API_KEY` |
| Default model | `glm-5.3` |
| Large context | `glm-5.3[1m]` |

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 108)May include surrounding context.

md
| Default model | `glm-5.3` |
| Large context | `glm-5.3[1m]` |
| Fast response | `glm-5.3-flash` |
| OpenAI-compatible URL | `https://api.z.ai/api/coding/paas/v4` |

The `[1m]` suffix is what unlocks the 1M-token window. Omitting it is
a documented setup mistake. Pair it with